{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.48742098448884263,
  "eval_steps": 800,
  "global_step": 12000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 8.097165991902834e-09,
      "loss": 3.5744,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.048582995951417e-08,
      "loss": 3.6858,
      "step": 5
    },
    {
      "epoch": 0.0,
      "learning_rate": 8.097165991902834e-08,
      "loss": 3.6628,
      "step": 10
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.214574898785425e-07,
      "loss": 3.3928,
      "step": 15
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.6194331983805668e-07,
      "loss": 3.0895,
      "step": 20
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.0242914979757083e-07,
      "loss": 2.6585,
      "step": 25
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.42914979757085e-07,
      "loss": 2.1214,
      "step": 30
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.8340080971659917e-07,
      "loss": 1.7237,
      "step": 35
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2388663967611335e-07,
      "loss": 1.2235,
      "step": 40
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.6437246963562754e-07,
      "loss": 1.1867,
      "step": 45
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.0485829959514166e-07,
      "loss": 1.1259,
      "step": 50
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4534412955465585e-07,
      "loss": 1.0245,
      "step": 55
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.8582995951417e-07,
      "loss": 0.977,
      "step": 60
    },
    {
      "epoch": 0.0,
      "learning_rate": 5.263157894736842e-07,
      "loss": 0.9554,
      "step": 65
    },
    {
      "epoch": 0.0,
      "learning_rate": 5.668016194331983e-07,
      "loss": 0.9017,
      "step": 70
    },
    {
      "epoch": 0.0,
      "learning_rate": 6.072874493927125e-07,
      "loss": 0.8987,
      "step": 75
    },
    {
      "epoch": 0.0,
      "learning_rate": 6.477732793522267e-07,
      "loss": 0.8863,
      "step": 80
    },
    {
      "epoch": 0.0,
      "learning_rate": 6.882591093117408e-07,
      "loss": 0.9425,
      "step": 85
    },
    {
      "epoch": 0.0,
      "learning_rate": 7.287449392712551e-07,
      "loss": 0.91,
      "step": 90
    },
    {
      "epoch": 0.0,
      "learning_rate": 7.692307692307693e-07,
      "loss": 0.848,
      "step": 95
    },
    {
      "epoch": 0.0,
      "learning_rate": 8.097165991902833e-07,
      "loss": 0.8213,
      "step": 100
    },
    {
      "epoch": 0.0,
      "learning_rate": 8.502024291497975e-07,
      "loss": 0.8434,
      "step": 105
    },
    {
      "epoch": 0.0,
      "learning_rate": 8.906882591093117e-07,
      "loss": 0.8409,
      "step": 110
    },
    {
      "epoch": 0.0,
      "learning_rate": 9.311740890688259e-07,
      "loss": 0.8398,
      "step": 115
    },
    {
      "epoch": 0.0,
      "learning_rate": 9.7165991902834e-07,
      "loss": 0.7942,
      "step": 120
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.0121457489878542e-06,
      "loss": 0.8221,
      "step": 125
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.0526315789473683e-06,
      "loss": 0.8038,
      "step": 130
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.0931174089068826e-06,
      "loss": 0.803,
      "step": 135
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.1336032388663967e-06,
      "loss": 0.7935,
      "step": 140
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.1740890688259108e-06,
      "loss": 0.8251,
      "step": 145
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.214574898785425e-06,
      "loss": 0.8082,
      "step": 150
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.2550607287449393e-06,
      "loss": 0.7897,
      "step": 155
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.2955465587044534e-06,
      "loss": 0.8286,
      "step": 160
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.3360323886639675e-06,
      "loss": 0.7742,
      "step": 165
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.3765182186234816e-06,
      "loss": 0.787,
      "step": 170
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.4170040485829959e-06,
      "loss": 0.7862,
      "step": 175
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.4574898785425101e-06,
      "loss": 0.7721,
      "step": 180
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.4979757085020242e-06,
      "loss": 0.7554,
      "step": 185
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.5384615384615385e-06,
      "loss": 0.7941,
      "step": 190
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.5789473684210526e-06,
      "loss": 0.7759,
      "step": 195
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.6194331983805667e-06,
      "loss": 0.7249,
      "step": 200
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.6599190283400807e-06,
      "loss": 0.7673,
      "step": 205
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.700404858299595e-06,
      "loss": 0.7922,
      "step": 210
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.7408906882591093e-06,
      "loss": 0.7546,
      "step": 215
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.7813765182186234e-06,
      "loss": 0.7709,
      "step": 220
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.8218623481781377e-06,
      "loss": 0.7383,
      "step": 225
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.8623481781376518e-06,
      "loss": 0.7608,
      "step": 230
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9028340080971658e-06,
      "loss": 0.7663,
      "step": 235
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.94331983805668e-06,
      "loss": 0.7567,
      "step": 240
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.983805668016194e-06,
      "loss": 0.751,
      "step": 245
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9999999252295637e-06,
      "loss": 0.8148,
      "step": 250
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9999994682991603e-06,
      "loss": 0.7634,
      "step": 255
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.999998595977674e-06,
      "loss": 0.7448,
      "step": 260
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.999997308265467e-06,
      "loss": 0.7508,
      "step": 265
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.999995605163075e-06,
      "loss": 0.7696,
      "step": 270
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9999934866712048e-06,
      "loss": 0.7676,
      "step": 275
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9999909527907367e-06,
      "loss": 0.7601,
      "step": 280
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9999880035227236e-06,
      "loss": 0.7779,
      "step": 285
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9999846388683895e-06,
      "loss": 0.7768,
      "step": 290
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9999808588291327e-06,
      "loss": 0.7713,
      "step": 295
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.999976663406524e-06,
      "loss": 0.7666,
      "step": 300
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.999972052602305e-06,
      "loss": 0.742,
      "step": 305
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.999967026418392e-06,
      "loss": 0.783,
      "step": 310
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.999961584856872e-06,
      "loss": 0.7269,
      "step": 315
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9999557279200056e-06,
      "loss": 0.7336,
      "step": 320
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9999494556102263e-06,
      "loss": 0.7072,
      "step": 325
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9999427679301387e-06,
      "loss": 0.7709,
      "step": 330
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.999935664882522e-06,
      "loss": 0.7237,
      "step": 335
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9999281464703247e-06,
      "loss": 0.719,
      "step": 340
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.999920212696672e-06,
      "loss": 0.748,
      "step": 345
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.999911863564859e-06,
      "loss": 0.7167,
      "step": 350
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9999030990783527e-06,
      "loss": 0.7151,
      "step": 355
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.999893919240795e-06,
      "loss": 0.7095,
      "step": 360
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9998843240559986e-06,
      "loss": 0.7703,
      "step": 365
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9998743135279497e-06,
      "loss": 0.7456,
      "step": 370
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.999863887660806e-06,
      "loss": 0.7532,
      "step": 375
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.999853046458899e-06,
      "loss": 0.7014,
      "step": 380
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9998417899267313e-06,
      "loss": 0.7629,
      "step": 385
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.999830118068979e-06,
      "loss": 0.7329,
      "step": 390
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.999818030890491e-06,
      "loss": 0.723,
      "step": 395
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.999805528396288e-06,
      "loss": 0.7549,
      "step": 400
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9997926105915627e-06,
      "loss": 0.7121,
      "step": 405
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.999779277481682e-06,
      "loss": 0.7506,
      "step": 410
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9997655290721834e-06,
      "loss": 0.7284,
      "step": 415
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9997513653687786e-06,
      "loss": 0.7344,
      "step": 420
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.999736786377351e-06,
      "loss": 0.7684,
      "step": 425
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9997217921039567e-06,
      "loss": 0.7427,
      "step": 430
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9997063825548237e-06,
      "loss": 0.7139,
      "step": 435
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9996905577363533e-06,
      "loss": 0.761,
      "step": 440
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9996743176551186e-06,
      "loss": 0.7545,
      "step": 445
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.999657662317866e-06,
      "loss": 0.7431,
      "step": 450
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.999640591731515e-06,
      "loss": 0.7225,
      "step": 455
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.999623105903154e-06,
      "loss": 0.7284,
      "step": 460
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.999605204840049e-06,
      "loss": 0.76,
      "step": 465
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9995868885496343e-06,
      "loss": 0.7413,
      "step": 470
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9995681570395195e-06,
      "loss": 0.7837,
      "step": 475
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9995490103174847e-06,
      "loss": 0.7347,
      "step": 480
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.999529448391483e-06,
      "loss": 0.7576,
      "step": 485
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9995094712696413e-06,
      "loss": 0.7665,
      "step": 490
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9994890789602576e-06,
      "loss": 0.7353,
      "step": 495
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.999468271471802e-06,
      "loss": 0.7344,
      "step": 500
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9994470488129185e-06,
      "loss": 0.7476,
      "step": 505
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9994254109924223e-06,
      "loss": 0.7257,
      "step": 510
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9994033580193017e-06,
      "loss": 0.7306,
      "step": 515
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.999380889902718e-06,
      "loss": 0.7115,
      "step": 520
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9993580066520034e-06,
      "loss": 0.7452,
      "step": 525
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9993347082766636e-06,
      "loss": 0.7523,
      "step": 530
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9993109947863764e-06,
      "loss": 0.7091,
      "step": 535
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.999286866190993e-06,
      "loss": 0.7383,
      "step": 540
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.999262322500535e-06,
      "loss": 0.7043,
      "step": 545
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9992373637251982e-06,
      "loss": 0.7098,
      "step": 550
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.999211989875351e-06,
      "loss": 0.7142,
      "step": 555
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.999186200961532e-06,
      "loss": 0.7424,
      "step": 560
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9991599969944552e-06,
      "loss": 0.7348,
      "step": 565
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9991333779850043e-06,
      "loss": 0.7126,
      "step": 570
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.999106343944237e-06,
      "loss": 0.7341,
      "step": 575
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9990788948833833e-06,
      "loss": 0.7445,
      "step": 580
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.999051030813845e-06,
      "loss": 0.7181,
      "step": 585
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.999022751747197e-06,
      "loss": 0.7295,
      "step": 590
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.998994057695185e-06,
      "loss": 0.7159,
      "step": 595
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.99896494866973e-06,
      "loss": 0.6844,
      "step": 600
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9989354246829222e-06,
      "loss": 0.7511,
      "step": 605
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9989054857470267e-06,
      "loss": 0.7322,
      "step": 610
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9988751318744787e-06,
      "loss": 0.7829,
      "step": 615
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.998844363077888e-06,
      "loss": 0.7229,
      "step": 620
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.998813179370035e-06,
      "loss": 0.738,
      "step": 625
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9987815807638733e-06,
      "loss": 0.6934,
      "step": 630
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9987495672725294e-06,
      "loss": 0.7005,
      "step": 635
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9987171389093e-06,
      "loss": 0.7692,
      "step": 640
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.998684295687657e-06,
      "loss": 0.7101,
      "step": 645
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.998651037621242e-06,
      "loss": 0.7813,
      "step": 650
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9986173647238715e-06,
      "loss": 0.7526,
      "step": 655
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9985832770095313e-06,
      "loss": 0.7235,
      "step": 660
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.998548774492382e-06,
      "loss": 0.7201,
      "step": 665
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9985138571867557e-06,
      "loss": 0.7303,
      "step": 670
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.998478525107157e-06,
      "loss": 0.7375,
      "step": 675
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.998442778268262e-06,
      "loss": 0.7123,
      "step": 680
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.99840661668492e-06,
      "loss": 0.7541,
      "step": 685
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.998370040372151e-06,
      "loss": 0.7685,
      "step": 690
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.99833304934515e-06,
      "loss": 0.7029,
      "step": 695
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9982956436192827e-06,
      "loss": 0.7797,
      "step": 700
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9982578232100866e-06,
      "loss": 0.7326,
      "step": 705
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9982195881332714e-06,
      "loss": 0.773,
      "step": 710
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9981809384047207e-06,
      "loss": 0.741,
      "step": 715
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9981418740404886e-06,
      "loss": 0.7518,
      "step": 720
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.998102395056802e-06,
      "loss": 0.7338,
      "step": 725
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.998062501470061e-06,
      "loss": 0.7192,
      "step": 730
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.998022193296836e-06,
      "loss": 0.7429,
      "step": 735
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9979814705538715e-06,
      "loss": 0.6953,
      "step": 740
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.997940333258083e-06,
      "loss": 0.7265,
      "step": 745
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9978987814265583e-06,
      "loss": 0.7105,
      "step": 750
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.997856815076558e-06,
      "loss": 0.6994,
      "step": 755
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9978144342255147e-06,
      "loss": 0.7008,
      "step": 760
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9977716388910325e-06,
      "loss": 0.7301,
      "step": 765
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.997728429090889e-06,
      "loss": 0.7662,
      "step": 770
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9976848048430323e-06,
      "loss": 0.7428,
      "step": 775
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9976407661655844e-06,
      "loss": 0.706,
      "step": 780
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.997596313076838e-06,
      "loss": 0.6853,
      "step": 785
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9975514455952584e-06,
      "loss": 0.7363,
      "step": 790
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9975061637394834e-06,
      "loss": 0.7217,
      "step": 795
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.997460467528323e-06,
      "loss": 0.7161,
      "step": 800
    },
    {
      "epoch": 0.03,
      "eval_loss": 0.6896045207977295,
      "eval_runtime": 140.4315,
      "eval_samples_per_second": 16.848,
      "eval_steps_per_second": 2.813,
      "step": 800
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.997414356980759e-06,
      "loss": 0.7911,
      "step": 805
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9973678321159443e-06,
      "loss": 0.7037,
      "step": 810
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9973208929532063e-06,
      "loss": 0.7083,
      "step": 815
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9972735395120418e-06,
      "loss": 0.7183,
      "step": 820
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.997225771812122e-06,
      "loss": 0.7227,
      "step": 825
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9971775898732893e-06,
      "loss": 0.7271,
      "step": 830
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9971289937155577e-06,
      "loss": 0.7271,
      "step": 835
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.997079983359113e-06,
      "loss": 0.7065,
      "step": 840
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9970305588243145e-06,
      "loss": 0.706,
      "step": 845
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9969807201316925e-06,
      "loss": 0.7,
      "step": 850
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9969304673019494e-06,
      "loss": 0.7165,
      "step": 855
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.99687980035596e-06,
      "loss": 0.729,
      "step": 860
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.996828719314771e-06,
      "loss": 0.7199,
      "step": 865
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.996777224199601e-06,
      "loss": 0.7041,
      "step": 870
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.99672531503184e-06,
      "loss": 0.735,
      "step": 875
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.996672991833051e-06,
      "loss": 0.7153,
      "step": 880
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.996620254624969e-06,
      "loss": 0.714,
      "step": 885
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9965671034295e-06,
      "loss": 0.7309,
      "step": 890
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.996513538268723e-06,
      "loss": 0.7808,
      "step": 895
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9964595591648883e-06,
      "loss": 0.7407,
      "step": 900
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9964051661404185e-06,
      "loss": 0.6831,
      "step": 905
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9963503592179078e-06,
      "loss": 0.7178,
      "step": 910
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.996295138420122e-06,
      "loss": 0.7607,
      "step": 915
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9962395037700007e-06,
      "loss": 0.747,
      "step": 920
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.996183455290653e-06,
      "loss": 0.6911,
      "step": 925
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.996126993005361e-06,
      "loss": 0.7038,
      "step": 930
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.996070116937579e-06,
      "loss": 0.7195,
      "step": 935
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9960128271109326e-06,
      "loss": 0.6974,
      "step": 940
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9959551235492195e-06,
      "loss": 0.7399,
      "step": 945
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9958970062764095e-06,
      "loss": 0.7475,
      "step": 950
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9958384753166437e-06,
      "loss": 0.7091,
      "step": 955
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.995779530694236e-06,
      "loss": 0.6908,
      "step": 960
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9957201724336704e-06,
      "loss": 0.7052,
      "step": 965
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9956604005596043e-06,
      "loss": 0.6963,
      "step": 970
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9956002150968667e-06,
      "loss": 0.7064,
      "step": 975
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9955396160704582e-06,
      "loss": 0.6804,
      "step": 980
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.99547860350555e-06,
      "loss": 0.6759,
      "step": 985
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.995417177427488e-06,
      "loss": 0.7175,
      "step": 990
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9953553378617866e-06,
      "loss": 0.6926,
      "step": 995
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.995293084834134e-06,
      "loss": 0.7109,
      "step": 1000
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9952304183703893e-06,
      "loss": 0.7129,
      "step": 1005
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9951673384965835e-06,
      "loss": 0.7117,
      "step": 1010
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.99510384523892e-06,
      "loss": 0.7694,
      "step": 1015
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.995039938623773e-06,
      "loss": 0.7381,
      "step": 1020
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9949756186776893e-06,
      "loss": 0.722,
      "step": 1025
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9949108854273855e-06,
      "loss": 0.7288,
      "step": 1030
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9948457388997528e-06,
      "loss": 0.7045,
      "step": 1035
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.994780179121851e-06,
      "loss": 0.7623,
      "step": 1040
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.994714206120914e-06,
      "loss": 0.725,
      "step": 1045
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9946478199243466e-06,
      "loss": 0.7203,
      "step": 1050
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9945810205597246e-06,
      "loss": 0.7011,
      "step": 1055
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9945138080547957e-06,
      "loss": 0.6946,
      "step": 1060
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.99444618243748e-06,
      "loss": 0.7151,
      "step": 1065
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.994378143735868e-06,
      "loss": 0.7074,
      "step": 1070
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9943096919782225e-06,
      "loss": 0.7,
      "step": 1075
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.994240827192978e-06,
      "loss": 0.6957,
      "step": 1080
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9941715494087408e-06,
      "loss": 0.7348,
      "step": 1085
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9941018586542866e-06,
      "loss": 0.6984,
      "step": 1090
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9940317549585665e-06,
      "loss": 0.7252,
      "step": 1095
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9939612383506993e-06,
      "loss": 0.7706,
      "step": 1100
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.993890308859978e-06,
      "loss": 0.7261,
      "step": 1105
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9938189665158654e-06,
      "loss": 0.6879,
      "step": 1110
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9937472113479966e-06,
      "loss": 0.7088,
      "step": 1115
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9936750433861787e-06,
      "loss": 0.7428,
      "step": 1120
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.993602462660389e-06,
      "loss": 0.7111,
      "step": 1125
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.993529469200777e-06,
      "loss": 0.7027,
      "step": 1130
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.993456063037664e-06,
      "loss": 0.6969,
      "step": 1135
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9933822442015416e-06,
      "loss": 0.7343,
      "step": 1140
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.993308012723074e-06,
      "loss": 0.7174,
      "step": 1145
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.993233368633096e-06,
      "loss": 0.6997,
      "step": 1150
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.993158311962614e-06,
      "loss": 0.693,
      "step": 1155
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9930828427428066e-06,
      "loss": 0.7136,
      "step": 1160
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9930069610050224e-06,
      "loss": 0.7211,
      "step": 1165
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9929306667807823e-06,
      "loss": 0.7144,
      "step": 1170
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.992853960101778e-06,
      "loss": 0.6787,
      "step": 1175
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9927768409998733e-06,
      "loss": 0.7348,
      "step": 1180
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.992699309507102e-06,
      "loss": 0.6718,
      "step": 1185
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.992621365655671e-06,
      "loss": 0.7146,
      "step": 1190
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9925430094779566e-06,
      "loss": 0.6982,
      "step": 1195
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9924642410065075e-06,
      "loss": 0.7379,
      "step": 1200
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.992385060274044e-06,
      "loss": 0.6983,
      "step": 1205
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9923054673134564e-06,
      "loss": 0.7893,
      "step": 1210
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.992225462157807e-06,
      "loss": 0.714,
      "step": 1215
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.99214504484033e-06,
      "loss": 0.7394,
      "step": 1220
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9920642153944288e-06,
      "loss": 0.7238,
      "step": 1225
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9919829738536806e-06,
      "loss": 0.6847,
      "step": 1230
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.991901320251831e-06,
      "loss": 0.6936,
      "step": 1235
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9918192546227995e-06,
      "loss": 0.7271,
      "step": 1240
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.991736777000675e-06,
      "loss": 0.7416,
      "step": 1245
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9916538874197176e-06,
      "loss": 0.7637,
      "step": 1250
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9915705859143594e-06,
      "loss": 0.6722,
      "step": 1255
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9914868725192025e-06,
      "loss": 0.6943,
      "step": 1260
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.991402747269022e-06,
      "loss": 0.7433,
      "step": 1265
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.991318210198761e-06,
      "loss": 0.7015,
      "step": 1270
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.991233261343537e-06,
      "loss": 0.6772,
      "step": 1275
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9911479007386364e-06,
      "loss": 0.7278,
      "step": 1280
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.991062128419517e-06,
      "loss": 0.7471,
      "step": 1285
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9909759444218085e-06,
      "loss": 0.7234,
      "step": 1290
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9908893487813106e-06,
      "loss": 0.7118,
      "step": 1295
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.990802341533994e-06,
      "loss": 0.7351,
      "step": 1300
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9907149227160016e-06,
      "loss": 0.7084,
      "step": 1305
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9906270923636457e-06,
      "loss": 0.7174,
      "step": 1310
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9905388505134107e-06,
      "loss": 0.6935,
      "step": 1315
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.990450197201951e-06,
      "loss": 0.7004,
      "step": 1320
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.990361132466093e-06,
      "loss": 0.7077,
      "step": 1325
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9902716563428335e-06,
      "loss": 0.7226,
      "step": 1330
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9901817688693395e-06,
      "loss": 0.7025,
      "step": 1335
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.99009147008295e-06,
      "loss": 0.7139,
      "step": 1340
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9900007600211735e-06,
      "loss": 0.6609,
      "step": 1345
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9899096387216914e-06,
      "loss": 0.7452,
      "step": 1350
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9898181062223536e-06,
      "loss": 0.7111,
      "step": 1355
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9897261625611822e-06,
      "loss": 0.6925,
      "step": 1360
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9896338077763704e-06,
      "loss": 0.7097,
      "step": 1365
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.989541041906281e-06,
      "loss": 0.7146,
      "step": 1370
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9894478649894484e-06,
      "loss": 0.704,
      "step": 1375
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.989354277064577e-06,
      "loss": 0.7466,
      "step": 1380
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9892602781705427e-06,
      "loss": 0.6958,
      "step": 1385
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9891658683463922e-06,
      "loss": 0.7421,
      "step": 1390
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.989071047631342e-06,
      "loss": 0.6658,
      "step": 1395
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.98897581606478e-06,
      "loss": 0.6868,
      "step": 1400
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.988880173686265e-06,
      "loss": 0.7437,
      "step": 1405
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.988784120535525e-06,
      "loss": 0.7484,
      "step": 1410
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.988687656652461e-06,
      "loss": 0.7063,
      "step": 1415
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9885907820771415e-06,
      "loss": 0.713,
      "step": 1420
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.988493496849809e-06,
      "loss": 0.7313,
      "step": 1425
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9883958010108736e-06,
      "loss": 0.6987,
      "step": 1430
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9882976946009186e-06,
      "loss": 0.7089,
      "step": 1435
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9881991776606956e-06,
      "loss": 0.6492,
      "step": 1440
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9881002502311285e-06,
      "loss": 0.6538,
      "step": 1445
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9880009123533095e-06,
      "loss": 0.7096,
      "step": 1450
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9879011640685043e-06,
      "loss": 0.7329,
      "step": 1455
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9878010054181463e-06,
      "loss": 0.7414,
      "step": 1460
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9877004364438414e-06,
      "loss": 0.7089,
      "step": 1465
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.987599457187365e-06,
      "loss": 0.738,
      "step": 1470
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9874980676906617e-06,
      "loss": 0.7078,
      "step": 1475
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9873962679958494e-06,
      "loss": 0.6987,
      "step": 1480
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.987294058145214e-06,
      "loss": 0.7456,
      "step": 1485
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.987191438181213e-06,
      "loss": 0.7402,
      "step": 1490
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.987088408146473e-06,
      "loss": 0.7308,
      "step": 1495
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.986984968083793e-06,
      "loss": 0.7197,
      "step": 1500
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9868811180361402e-06,
      "loss": 0.7386,
      "step": 1505
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9867768580466536e-06,
      "loss": 0.7024,
      "step": 1510
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.986672188158641e-06,
      "loss": 0.7241,
      "step": 1515
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9865671084155826e-06,
      "loss": 0.7006,
      "step": 1520
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.986461618861127e-06,
      "loss": 0.7035,
      "step": 1525
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.986355719539093e-06,
      "loss": 0.723,
      "step": 1530
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9862494104934717e-06,
      "loss": 0.7184,
      "step": 1535
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9861426917684214e-06,
      "loss": 0.7018,
      "step": 1540
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.986035563408273e-06,
      "loss": 0.6943,
      "step": 1545
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9859280254575268e-06,
      "loss": 0.7434,
      "step": 1550
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9858200779608526e-06,
      "loss": 0.7122,
      "step": 1555
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9857117209630913e-06,
      "loss": 0.7187,
      "step": 1560
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9856029545092536e-06,
      "loss": 0.6825,
      "step": 1565
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.985493778644519e-06,
      "loss": 0.6964,
      "step": 1570
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9853841934142396e-06,
      "loss": 0.7437,
      "step": 1575
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9852741988639356e-06,
      "loss": 0.7125,
      "step": 1580
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9851637950392974e-06,
      "loss": 0.7241,
      "step": 1585
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9850529819861863e-06,
      "loss": 0.7113,
      "step": 1590
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.984941759750633e-06,
      "loss": 0.6725,
      "step": 1595
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.984830128378838e-06,
      "loss": 0.7166,
      "step": 1600
    },
    {
      "epoch": 0.06,
      "eval_loss": 0.6776626706123352,
      "eval_runtime": 140.3492,
      "eval_samples_per_second": 16.858,
      "eval_steps_per_second": 2.814,
      "step": 1600
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9847180879171727e-06,
      "loss": 0.7111,
      "step": 1605
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9846056384121768e-06,
      "loss": 0.7004,
      "step": 1610
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9844927799105612e-06,
      "loss": 0.7221,
      "step": 1615
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.984379512459207e-06,
      "loss": 0.7363,
      "step": 1620
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.984265836105163e-06,
      "loss": 0.7107,
      "step": 1625
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9841517508956506e-06,
      "loss": 0.7081,
      "step": 1630
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9840372568780594e-06,
      "loss": 0.6796,
      "step": 1635
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9839223540999496e-06,
      "loss": 0.7207,
      "step": 1640
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9838070426090505e-06,
      "loss": 0.716,
      "step": 1645
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.983691322453261e-06,
      "loss": 0.7306,
      "step": 1650
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.983575193680651e-06,
      "loss": 0.724,
      "step": 1655
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.983458656339459e-06,
      "loss": 0.7447,
      "step": 1660
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9833417104780942e-06,
      "loss": 0.6929,
      "step": 1665
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9832243561451346e-06,
      "loss": 0.7228,
      "step": 1670
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9831065933893275e-06,
      "loss": 0.6824,
      "step": 1675
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.982988422259591e-06,
      "loss": 0.7056,
      "step": 1680
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9828698428050123e-06,
      "loss": 0.6943,
      "step": 1685
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.982750855074849e-06,
      "loss": 0.7101,
      "step": 1690
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9826314591185263e-06,
      "loss": 0.6786,
      "step": 1695
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9825116549856408e-06,
      "loss": 0.6954,
      "step": 1700
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9823914427259584e-06,
      "loss": 0.7165,
      "step": 1705
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.982270822389414e-06,
      "loss": 0.7208,
      "step": 1710
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9821497940261124e-06,
      "loss": 0.6981,
      "step": 1715
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.982028357686327e-06,
      "loss": 0.6914,
      "step": 1720
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9819065134205026e-06,
      "loss": 0.7291,
      "step": 1725
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9817842612792513e-06,
      "loss": 0.6882,
      "step": 1730
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.981661601313356e-06,
      "loss": 0.685,
      "step": 1735
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.981538533573768e-06,
      "loss": 0.6954,
      "step": 1740
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9814150581116093e-06,
      "loss": 0.7104,
      "step": 1745
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9812911749781705e-06,
      "loss": 0.7026,
      "step": 1750
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.981166884224911e-06,
      "loss": 0.6907,
      "step": 1755
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.981042185903461e-06,
      "loss": 0.6988,
      "step": 1760
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.980917080065618e-06,
      "loss": 0.6894,
      "step": 1765
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.98079156676335e-06,
      "loss": 0.7308,
      "step": 1770
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9806656460487955e-06,
      "loss": 0.6688,
      "step": 1775
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9805393179742596e-06,
      "loss": 0.7028,
      "step": 1780
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.980412582592218e-06,
      "loss": 0.6982,
      "step": 1785
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.980285439955316e-06,
      "loss": 0.7326,
      "step": 1790
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.980157890116367e-06,
      "loss": 0.7204,
      "step": 1795
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.980029933128354e-06,
      "loss": 0.7016,
      "step": 1800
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9799015690444302e-06,
      "loss": 0.7076,
      "step": 1805
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9797727979179156e-06,
      "loss": 0.7121,
      "step": 1810
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9796436198023016e-06,
      "loss": 0.7204,
      "step": 1815
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9795140347512472e-06,
      "loss": 0.7178,
      "step": 1820
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.979384042818581e-06,
      "loss": 0.7223,
      "step": 1825
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.979253644058301e-06,
      "loss": 0.7066,
      "step": 1830
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.979122838524573e-06,
      "loss": 0.6873,
      "step": 1835
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9789916262717328e-06,
      "loss": 0.6822,
      "step": 1840
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9788600073542848e-06,
      "loss": 0.6947,
      "step": 1845
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.978727981826902e-06,
      "loss": 0.7092,
      "step": 1850
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.978595549744427e-06,
      "loss": 0.7166,
      "step": 1855
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9784627111618715e-06,
      "loss": 0.6842,
      "step": 1860
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9783294661344145e-06,
      "loss": 0.7161,
      "step": 1865
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.978195814717405e-06,
      "loss": 0.6881,
      "step": 1870
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.978061756966361e-06,
      "loss": 0.7342,
      "step": 1875
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.977927292936969e-06,
      "loss": 0.6767,
      "step": 1880
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9777924226850842e-06,
      "loss": 0.7096,
      "step": 1885
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.97765714626673e-06,
      "loss": 0.694,
      "step": 1890
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.977521463738099e-06,
      "loss": 0.7152,
      "step": 1895
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9773853751555537e-06,
      "loss": 0.6618,
      "step": 1900
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.977248880575623e-06,
      "loss": 0.689,
      "step": 1905
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9771119800550054e-06,
      "loss": 0.6892,
      "step": 1910
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9769746736505694e-06,
      "loss": 0.7179,
      "step": 1915
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.97683696141935e-06,
      "loss": 0.6888,
      "step": 1920
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9766988434185514e-06,
      "loss": 0.7041,
      "step": 1925
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.976560319705547e-06,
      "loss": 0.6969,
      "step": 1930
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9764213903378786e-06,
      "loss": 0.7162,
      "step": 1935
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9762820553732563e-06,
      "loss": 0.7178,
      "step": 1940
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.976142314869558e-06,
      "loss": 0.7309,
      "step": 1945
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.976002168884831e-06,
      "loss": 0.7198,
      "step": 1950
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.975861617477291e-06,
      "loss": 0.7131,
      "step": 1955
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9757206607053218e-06,
      "loss": 0.7087,
      "step": 1960
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9755792986274755e-06,
      "loss": 0.6708,
      "step": 1965
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.975437531302472e-06,
      "loss": 0.7141,
      "step": 1970
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.975295358789201e-06,
      "loss": 0.7152,
      "step": 1975
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9751527811467195e-06,
      "loss": 0.7172,
      "step": 1980
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9750097984342534e-06,
      "loss": 0.7472,
      "step": 1985
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9748664107111962e-06,
      "loss": 0.7129,
      "step": 1990
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9747226180371094e-06,
      "loss": 0.7066,
      "step": 1995
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.974578420471724e-06,
      "loss": 0.7049,
      "step": 2000
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9744338180749376e-06,
      "loss": 0.7214,
      "step": 2005
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9742888109068175e-06,
      "loss": 0.7469,
      "step": 2010
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9741433990275987e-06,
      "loss": 0.7119,
      "step": 2015
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.973997582497683e-06,
      "loss": 0.7487,
      "step": 2020
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.973851361377642e-06,
      "loss": 0.7259,
      "step": 2025
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9737047357282143e-06,
      "loss": 0.7164,
      "step": 2030
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9735577056103074e-06,
      "loss": 0.7639,
      "step": 2035
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9734102710849956e-06,
      "loss": 0.7181,
      "step": 2040
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.973262432213523e-06,
      "loss": 0.6833,
      "step": 2045
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.973114189057299e-06,
      "loss": 0.7263,
      "step": 2050
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9729655416779044e-06,
      "loss": 0.6957,
      "step": 2055
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.972816490137085e-06,
      "loss": 0.716,
      "step": 2060
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9726670344967554e-06,
      "loss": 0.699,
      "step": 2065
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9725171748189987e-06,
      "loss": 0.6858,
      "step": 2070
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9723669111660645e-06,
      "loss": 0.7036,
      "step": 2075
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9722162436003715e-06,
      "loss": 0.6958,
      "step": 2080
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9720651721845062e-06,
      "loss": 0.7303,
      "step": 2085
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.971913696981222e-06,
      "loss": 0.6836,
      "step": 2090
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.97176181805344e-06,
      "loss": 0.7349,
      "step": 2095
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9716095354642493e-06,
      "loss": 0.7105,
      "step": 2100
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.971456849276907e-06,
      "loss": 0.7126,
      "step": 2105
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.971303759554838e-06,
      "loss": 0.6959,
      "step": 2110
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9711502663616344e-06,
      "loss": 0.6699,
      "step": 2115
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9709963697610554e-06,
      "loss": 0.6671,
      "step": 2120
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.970842069817029e-06,
      "loss": 0.7101,
      "step": 2125
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.970687366593649e-06,
      "loss": 0.6985,
      "step": 2130
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.970532260155179e-06,
      "loss": 0.7175,
      "step": 2135
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.970376750566048e-06,
      "loss": 0.7251,
      "step": 2140
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9702208378908537e-06,
      "loss": 0.6949,
      "step": 2145
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.970064522194361e-06,
      "loss": 0.7573,
      "step": 2150
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9699078035415014e-06,
      "loss": 0.6922,
      "step": 2155
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9697506819973753e-06,
      "loss": 0.7215,
      "step": 2160
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9695931576272493e-06,
      "loss": 0.7252,
      "step": 2165
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9694352304965572e-06,
      "loss": 0.7202,
      "step": 2170
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9692769006709013e-06,
      "loss": 0.6932,
      "step": 2175
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.96911816821605e-06,
      "loss": 0.6522,
      "step": 2180
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9689590331979394e-06,
      "loss": 0.6933,
      "step": 2185
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.968799495682673e-06,
      "loss": 0.6825,
      "step": 2190
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.968639555736521e-06,
      "loss": 0.7037,
      "step": 2195
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.968479213425922e-06,
      "loss": 0.6911,
      "step": 2200
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9683184688174795e-06,
      "loss": 0.7048,
      "step": 2205
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9681573219779658e-06,
      "loss": 0.7273,
      "step": 2210
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9679957729743204e-06,
      "loss": 0.7105,
      "step": 2215
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9678338218736486e-06,
      "loss": 0.6832,
      "step": 2220
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.967671468743224e-06,
      "loss": 0.6829,
      "step": 2225
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9675087136504865e-06,
      "loss": 0.6971,
      "step": 2230
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9673455566630437e-06,
      "loss": 0.7298,
      "step": 2235
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9671819978486688e-06,
      "loss": 0.6841,
      "step": 2240
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.967018037275303e-06,
      "loss": 0.6838,
      "step": 2245
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.966853675011054e-06,
      "loss": 0.7053,
      "step": 2250
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.966688911124197e-06,
      "loss": 0.6886,
      "step": 2255
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9665237456831725e-06,
      "loss": 0.7032,
      "step": 2260
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9663581787565898e-06,
      "loss": 0.7159,
      "step": 2265
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.966192210413223e-06,
      "loss": 0.6896,
      "step": 2270
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.966025840722015e-06,
      "loss": 0.7004,
      "step": 2275
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9658590697520735e-06,
      "loss": 0.6726,
      "step": 2280
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.965691897572674e-06,
      "loss": 0.6911,
      "step": 2285
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9655243242532584e-06,
      "loss": 0.7235,
      "step": 2290
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9653563498634347e-06,
      "loss": 0.7014,
      "step": 2295
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9651879744729786e-06,
      "loss": 0.7192,
      "step": 2300
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9650191981518317e-06,
      "loss": 0.65,
      "step": 2305
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.964850020970102e-06,
      "loss": 0.7128,
      "step": 2310
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9646804429980644e-06,
      "loss": 0.7265,
      "step": 2315
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.96451046430616e-06,
      "loss": 0.7038,
      "step": 2320
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.964340084964997e-06,
      "loss": 0.7132,
      "step": 2325
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.964169305045348e-06,
      "loss": 0.6968,
      "step": 2330
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9639981246181555e-06,
      "loss": 0.7456,
      "step": 2335
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9638265437545247e-06,
      "loss": 0.7403,
      "step": 2340
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9636545625257297e-06,
      "loss": 0.6893,
      "step": 2345
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.96348218100321e-06,
      "loss": 0.7341,
      "step": 2350
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9633093992585706e-06,
      "loss": 0.7459,
      "step": 2355
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.963136217363585e-06,
      "loss": 0.716,
      "step": 2360
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9629626353901897e-06,
      "loss": 0.7086,
      "step": 2365
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9627886534104903e-06,
      "loss": 0.7041,
      "step": 2370
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.962614271496757e-06,
      "loss": 0.7094,
      "step": 2375
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.962439489721427e-06,
      "loss": 0.6794,
      "step": 2380
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.962264308157102e-06,
      "loss": 0.6751,
      "step": 2385
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9620887268765523e-06,
      "loss": 0.7398,
      "step": 2390
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.961912745952712e-06,
      "loss": 0.6622,
      "step": 2395
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.961736365458682e-06,
      "loss": 0.7404,
      "step": 2400
    },
    {
      "epoch": 0.1,
      "eval_loss": 0.6729753613471985,
      "eval_runtime": 140.5864,
      "eval_samples_per_second": 16.83,
      "eval_steps_per_second": 2.81,
      "step": 2400
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.96155958546773e-06,
      "loss": 0.6928,
      "step": 2405
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.961382406053288e-06,
      "loss": 0.6832,
      "step": 2410
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.961204827288955e-06,
      "loss": 0.6775,
      "step": 2415
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.961026849248496e-06,
      "loss": 0.6995,
      "step": 2420
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9608484720058416e-06,
      "loss": 0.6508,
      "step": 2425
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.960669695635087e-06,
      "loss": 0.6865,
      "step": 2430
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.960490520210496e-06,
      "loss": 0.7273,
      "step": 2435
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9603109458064955e-06,
      "loss": 0.6513,
      "step": 2440
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9601309724976795e-06,
      "loss": 0.7051,
      "step": 2445
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9599506003588068e-06,
      "loss": 0.7164,
      "step": 2450
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9597698294648034e-06,
      "loss": 0.7269,
      "step": 2455
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.959588659890759e-06,
      "loss": 0.7082,
      "step": 2460
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9594070917119306e-06,
      "loss": 0.7164,
      "step": 2465
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9592251250037394e-06,
      "loss": 0.7156,
      "step": 2470
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9590427598417733e-06,
      "loss": 0.6603,
      "step": 2475
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.958859996301785e-06,
      "loss": 0.6811,
      "step": 2480
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.958676834459693e-06,
      "loss": 0.701,
      "step": 2485
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9584932743915807e-06,
      "loss": 0.6546,
      "step": 2490
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9583093161736975e-06,
      "loss": 0.7022,
      "step": 2495
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9581249598824588e-06,
      "loss": 0.7066,
      "step": 2500
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.957940205594444e-06,
      "loss": 0.6942,
      "step": 2505
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.957755053386398e-06,
      "loss": 0.6983,
      "step": 2510
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.957569503335232e-06,
      "loss": 0.7491,
      "step": 2515
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.957383555518022e-06,
      "loss": 0.6916,
      "step": 2520
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9571972100120087e-06,
      "loss": 0.7035,
      "step": 2525
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9570104668945986e-06,
      "loss": 0.6987,
      "step": 2530
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.956823326243363e-06,
      "loss": 0.7035,
      "step": 2535
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.956635788136039e-06,
      "loss": 0.6854,
      "step": 2540
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9564478526505276e-06,
      "loss": 0.7194,
      "step": 2545
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9562595198648956e-06,
      "loss": 0.7026,
      "step": 2550
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9560707898573757e-06,
      "loss": 0.6887,
      "step": 2555
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9558816627063636e-06,
      "loss": 0.7127,
      "step": 2560
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9556921384904217e-06,
      "loss": 0.7126,
      "step": 2565
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9555022172882767e-06,
      "loss": 0.671,
      "step": 2570
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9553118991788203e-06,
      "loss": 0.7004,
      "step": 2575
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9551211842411083e-06,
      "loss": 0.7303,
      "step": 2580
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9549300725543627e-06,
      "loss": 0.7571,
      "step": 2585
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9547385641979696e-06,
      "loss": 0.6688,
      "step": 2590
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9545466592514795e-06,
      "loss": 0.7083,
      "step": 2595
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9543543577946086e-06,
      "loss": 0.7426,
      "step": 2600
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9541616599072365e-06,
      "loss": 0.6722,
      "step": 2605
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.953968565669409e-06,
      "loss": 0.7215,
      "step": 2610
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9537750751613354e-06,
      "loss": 0.7354,
      "step": 2615
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.95358118846339e-06,
      "loss": 0.7253,
      "step": 2620
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9533869056561113e-06,
      "loss": 0.6877,
      "step": 2625
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9531922268202033e-06,
      "loss": 0.72,
      "step": 2630
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.952997152036534e-06,
      "loss": 0.6894,
      "step": 2635
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.952801681386135e-06,
      "loss": 0.69,
      "step": 2640
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9526058149502035e-06,
      "loss": 0.689,
      "step": 2645
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.952409552810101e-06,
      "loss": 0.7495,
      "step": 2650
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9522128950473525e-06,
      "loss": 0.7378,
      "step": 2655
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9520158417436486e-06,
      "loss": 0.7041,
      "step": 2660
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9518183929808433e-06,
      "loss": 0.711,
      "step": 2665
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9516205488409543e-06,
      "loss": 0.6908,
      "step": 2670
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9514223094061654e-06,
      "loss": 0.7107,
      "step": 2675
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.951223674758823e-06,
      "loss": 0.7225,
      "step": 2680
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9510246449814385e-06,
      "loss": 0.7217,
      "step": 2685
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.950825220156687e-06,
      "loss": 0.7274,
      "step": 2690
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.950625400367408e-06,
      "loss": 0.6867,
      "step": 2695
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9504251856966043e-06,
      "loss": 0.7186,
      "step": 2700
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.950224576227444e-06,
      "loss": 0.6889,
      "step": 2705
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9500235720432583e-06,
      "loss": 0.7005,
      "step": 2710
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9498221732275425e-06,
      "loss": 0.6874,
      "step": 2715
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9496203798639565e-06,
      "loss": 0.7033,
      "step": 2720
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9494181920363223e-06,
      "loss": 0.707,
      "step": 2725
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9492156098286283e-06,
      "loss": 0.7348,
      "step": 2730
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.949012633325024e-06,
      "loss": 0.6636,
      "step": 2735
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9488092626098254e-06,
      "loss": 0.6975,
      "step": 2740
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9486054977675103e-06,
      "loss": 0.7124,
      "step": 2745
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9484013388827204e-06,
      "loss": 0.663,
      "step": 2750
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9481967860402627e-06,
      "loss": 0.6916,
      "step": 2755
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9479918393251056e-06,
      "loss": 0.7344,
      "step": 2760
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9477864988223828e-06,
      "loss": 0.6752,
      "step": 2765
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9475807646173908e-06,
      "loss": 0.6885,
      "step": 2770
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9473746367955893e-06,
      "loss": 0.6602,
      "step": 2775
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.947168115442603e-06,
      "loss": 0.6764,
      "step": 2780
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.946961200644218e-06,
      "loss": 0.6784,
      "step": 2785
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.946753892486386e-06,
      "loss": 0.7325,
      "step": 2790
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.94654619105522e-06,
      "loss": 0.6533,
      "step": 2795
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.946338096436998e-06,
      "loss": 0.6741,
      "step": 2800
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9461296087181604e-06,
      "loss": 0.7046,
      "step": 2805
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9459207279853113e-06,
      "loss": 0.6797,
      "step": 2810
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.945711454325218e-06,
      "loss": 0.7096,
      "step": 2815
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9455017878248107e-06,
      "loss": 0.698,
      "step": 2820
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9452917285711834e-06,
      "loss": 0.7036,
      "step": 2825
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.945081276651593e-06,
      "loss": 0.7205,
      "step": 2830
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9448704321534588e-06,
      "loss": 0.6583,
      "step": 2835
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.944659195164364e-06,
      "loss": 0.6757,
      "step": 2840
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9444475657720545e-06,
      "loss": 0.7033,
      "step": 2845
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9442355440644394e-06,
      "loss": 0.7029,
      "step": 2850
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.944023130129591e-06,
      "loss": 0.6993,
      "step": 2855
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9438103240557446e-06,
      "loss": 0.704,
      "step": 2860
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9435971259312966e-06,
      "loss": 0.7104,
      "step": 2865
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9433835358448086e-06,
      "loss": 0.7017,
      "step": 2870
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.943169553885004e-06,
      "loss": 0.7107,
      "step": 2875
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9429551801407687e-06,
      "loss": 0.6939,
      "step": 2880
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.942740414701152e-06,
      "loss": 0.6737,
      "step": 2885
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9425252576553656e-06,
      "loss": 0.7174,
      "step": 2890
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.942309709092784e-06,
      "loss": 0.7147,
      "step": 2895
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9420937691029435e-06,
      "loss": 0.7351,
      "step": 2900
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9418774377755444e-06,
      "loss": 0.6845,
      "step": 2905
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9416607152004485e-06,
      "loss": 0.7156,
      "step": 2910
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9414436014676806e-06,
      "loss": 0.7532,
      "step": 2915
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.941226096667428e-06,
      "loss": 0.7421,
      "step": 2920
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.94100820089004e-06,
      "loss": 0.7049,
      "step": 2925
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.940789914226029e-06,
      "loss": 0.7652,
      "step": 2930
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.940571236766069e-06,
      "loss": 0.6884,
      "step": 2935
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9403521686009964e-06,
      "loss": 0.6974,
      "step": 2940
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.940132709821811e-06,
      "loss": 0.7362,
      "step": 2945
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9399128605196737e-06,
      "loss": 0.7167,
      "step": 2950
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9396926207859082e-06,
      "loss": 0.6806,
      "step": 2955
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.939471990712e-06,
      "loss": 0.7069,
      "step": 2960
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.939250970389597e-06,
      "loss": 0.7185,
      "step": 2965
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9390295599105085e-06,
      "loss": 0.6996,
      "step": 2970
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9388077593667075e-06,
      "loss": 0.7304,
      "step": 2975
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9385855688503276e-06,
      "loss": 0.7209,
      "step": 2980
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9383629884536644e-06,
      "loss": 0.7077,
      "step": 2985
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.938140018269176e-06,
      "loss": 0.6838,
      "step": 2990
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.937916658389483e-06,
      "loss": 0.6784,
      "step": 2995
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9376929089073665e-06,
      "loss": 0.7185,
      "step": 3000
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9374687699157703e-06,
      "loss": 0.7146,
      "step": 3005
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9372442415077994e-06,
      "loss": 0.7131,
      "step": 3010
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9370193237767213e-06,
      "loss": 0.7014,
      "step": 3015
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9367940168159648e-06,
      "loss": 0.697,
      "step": 3020
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9365683207191205e-06,
      "loss": 0.7009,
      "step": 3025
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9363422355799406e-06,
      "loss": 0.7125,
      "step": 3030
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.936115761492339e-06,
      "loss": 0.6954,
      "step": 3035
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.935888898550391e-06,
      "loss": 0.7465,
      "step": 3040
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.935661646848333e-06,
      "loss": 0.7253,
      "step": 3045
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.935434006480564e-06,
      "loss": 0.7223,
      "step": 3050
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.935205977541644e-06,
      "loss": 0.7141,
      "step": 3055
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9349775601262935e-06,
      "loss": 0.732,
      "step": 3060
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9347487543293958e-06,
      "loss": 0.709,
      "step": 3065
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.934519560245994e-06,
      "loss": 0.724,
      "step": 3070
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9342899779712946e-06,
      "loss": 0.6998,
      "step": 3075
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.934060007600663e-06,
      "loss": 0.6945,
      "step": 3080
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9338296492296267e-06,
      "loss": 0.7068,
      "step": 3085
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9335989029538756e-06,
      "loss": 0.691,
      "step": 3090
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9333677688692595e-06,
      "loss": 0.7063,
      "step": 3095
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9331362470717886e-06,
      "loss": 0.7518,
      "step": 3100
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9329043376576357e-06,
      "loss": 0.6806,
      "step": 3105
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.932672040723134e-06,
      "loss": 0.6878,
      "step": 3110
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9324393563647772e-06,
      "loss": 0.6961,
      "step": 3115
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.932206284679221e-06,
      "loss": 0.7125,
      "step": 3120
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.931972825763281e-06,
      "loss": 0.6893,
      "step": 3125
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.931738979713934e-06,
      "loss": 0.6821,
      "step": 3130
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9315047466283177e-06,
      "loss": 0.735,
      "step": 3135
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9312701266037302e-06,
      "loss": 0.6873,
      "step": 3140
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.931035119737631e-06,
      "loss": 0.6853,
      "step": 3145
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9307997261276393e-06,
      "loss": 0.7256,
      "step": 3150
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9305639458715365e-06,
      "loss": 0.7297,
      "step": 3155
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.930327779067263e-06,
      "loss": 0.7162,
      "step": 3160
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9300912258129206e-06,
      "loss": 0.7133,
      "step": 3165
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9298542862067712e-06,
      "loss": 0.7255,
      "step": 3170
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9296169603472384e-06,
      "loss": 0.6895,
      "step": 3175
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.929379248332904e-06,
      "loss": 0.729,
      "step": 3180
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9291411502625123e-06,
      "loss": 0.6971,
      "step": 3185
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.928902666234967e-06,
      "loss": 0.7232,
      "step": 3190
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9286637963493323e-06,
      "loss": 0.6653,
      "step": 3195
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9284245407048323e-06,
      "loss": 0.7309,
      "step": 3200
    },
    {
      "epoch": 0.13,
      "eval_loss": 0.6701433062553406,
      "eval_runtime": 140.1425,
      "eval_samples_per_second": 16.883,
      "eval_steps_per_second": 2.819,
      "step": 3200
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.928184899400853e-06,
      "loss": 0.7025,
      "step": 3205
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9279448725369375e-06,
      "loss": 0.711,
      "step": 3210
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.927704460212792e-06,
      "loss": 0.6813,
      "step": 3215
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9274636625282816e-06,
      "loss": 0.7209,
      "step": 3220
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.927222479583431e-06,
      "loss": 0.7279,
      "step": 3225
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9269809114784265e-06,
      "loss": 0.6779,
      "step": 3230
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.926738958313612e-06,
      "loss": 0.7237,
      "step": 3235
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9264966201894945e-06,
      "loss": 0.7123,
      "step": 3240
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9262538972067375e-06,
      "loss": 0.7147,
      "step": 3245
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9260107894661666e-06,
      "loss": 0.7104,
      "step": 3250
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9257672970687673e-06,
      "loss": 0.7059,
      "step": 3255
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9255234201156834e-06,
      "loss": 0.7259,
      "step": 3260
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9252791587082195e-06,
      "loss": 0.6778,
      "step": 3265
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9250345129478396e-06,
      "loss": 0.7456,
      "step": 3270
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.924789482936168e-06,
      "loss": 0.7254,
      "step": 3275
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.924544068774987e-06,
      "loss": 0.732,
      "step": 3280
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.92429827056624e-06,
      "loss": 0.6824,
      "step": 3285
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9240520884120296e-06,
      "loss": 0.6807,
      "step": 3290
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.923805522414618e-06,
      "loss": 0.6679,
      "step": 3295
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.923558572676426e-06,
      "loss": 0.6813,
      "step": 3300
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9233112393000344e-06,
      "loss": 0.6791,
      "step": 3305
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9230635223881836e-06,
      "loss": 0.6877,
      "step": 3310
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9228154220437733e-06,
      "loss": 0.7023,
      "step": 3315
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.922566938369861e-06,
      "loss": 0.7001,
      "step": 3320
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9223180714696664e-06,
      "loss": 0.691,
      "step": 3325
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.922068821446565e-06,
      "loss": 0.7149,
      "step": 3330
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9218191884040945e-06,
      "loss": 0.6826,
      "step": 3335
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9215691724459496e-06,
      "loss": 0.7069,
      "step": 3340
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9213187736759848e-06,
      "loss": 0.7044,
      "step": 3345
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9210679921982134e-06,
      "loss": 0.7026,
      "step": 3350
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9208168281168083e-06,
      "loss": 0.7151,
      "step": 3355
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9205652815361003e-06,
      "loss": 0.7147,
      "step": 3360
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.92031335256058e-06,
      "loss": 0.6891,
      "step": 3365
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9200610412948967e-06,
      "loss": 0.7246,
      "step": 3370
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9198083478438584e-06,
      "loss": 0.6832,
      "step": 3375
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.919555272312431e-06,
      "loss": 0.6708,
      "step": 3380
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.91930181480574e-06,
      "loss": 0.7217,
      "step": 3385
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9190479754290703e-06,
      "loss": 0.7272,
      "step": 3390
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.918793754287864e-06,
      "loss": 0.6629,
      "step": 3395
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.918539151487722e-06,
      "loss": 0.6633,
      "step": 3400
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9182841671344053e-06,
      "loss": 0.7085,
      "step": 3405
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.918028801333831e-06,
      "loss": 0.7321,
      "step": 3410
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9177730541920757e-06,
      "loss": 0.7141,
      "step": 3415
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9175169258153752e-06,
      "loss": 0.6976,
      "step": 3420
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9172604163101227e-06,
      "loss": 0.7034,
      "step": 3425
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9170035257828706e-06,
      "loss": 0.6478,
      "step": 3430
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9167462543403286e-06,
      "loss": 0.6949,
      "step": 3435
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9164886020893647e-06,
      "loss": 0.7224,
      "step": 3440
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9162305691370057e-06,
      "loss": 0.7016,
      "step": 3445
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9159721555904364e-06,
      "loss": 0.7065,
      "step": 3450
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9157133615569993e-06,
      "loss": 0.6808,
      "step": 3455
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9154541871441947e-06,
      "loss": 0.7026,
      "step": 3460
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9151946324596826e-06,
      "loss": 0.7035,
      "step": 3465
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9149346976112787e-06,
      "loss": 0.6941,
      "step": 3470
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9146743827069584e-06,
      "loss": 0.6988,
      "step": 3475
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9144136878548536e-06,
      "loss": 0.6812,
      "step": 3480
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.914152613163255e-06,
      "loss": 0.6976,
      "step": 3485
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.913891158740611e-06,
      "loss": 0.7263,
      "step": 3490
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.913629324695527e-06,
      "loss": 0.6872,
      "step": 3495
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.913367111136767e-06,
      "loss": 0.6588,
      "step": 3500
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.913104518173252e-06,
      "loss": 0.6679,
      "step": 3505
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.912841545914061e-06,
      "loss": 0.7026,
      "step": 3510
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9125781944684304e-06,
      "loss": 0.6763,
      "step": 3515
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9123144639457537e-06,
      "loss": 0.7106,
      "step": 3520
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9120503544555826e-06,
      "loss": 0.7148,
      "step": 3525
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.911785866107626e-06,
      "loss": 0.7029,
      "step": 3530
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9115209990117495e-06,
      "loss": 0.7071,
      "step": 3535
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9112557532779773e-06,
      "loss": 0.6967,
      "step": 3540
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9109901290164897e-06,
      "loss": 0.6872,
      "step": 3545
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9107241263376255e-06,
      "loss": 0.6732,
      "step": 3550
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9104577453518783e-06,
      "loss": 0.7015,
      "step": 3555
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.910190986169902e-06,
      "loss": 0.6894,
      "step": 3560
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9099238489025054e-06,
      "loss": 0.6748,
      "step": 3565
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.909656333660655e-06,
      "loss": 0.6691,
      "step": 3570
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.909388440555474e-06,
      "loss": 0.6763,
      "step": 3575
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.909120169698244e-06,
      "loss": 0.7212,
      "step": 3580
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9088515212004006e-06,
      "loss": 0.6999,
      "step": 3585
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.90858249517354e-06,
      "loss": 0.7089,
      "step": 3590
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9083130917294116e-06,
      "loss": 0.6819,
      "step": 3595
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9080433109799243e-06,
      "loss": 0.6902,
      "step": 3600
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9077731530371425e-06,
      "loss": 0.6825,
      "step": 3605
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9075026180132873e-06,
      "loss": 0.7008,
      "step": 3610
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9072317060207364e-06,
      "loss": 0.6789,
      "step": 3615
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9069604171720243e-06,
      "loss": 0.7221,
      "step": 3620
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9066887515798426e-06,
      "loss": 0.6761,
      "step": 3625
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9064167093570382e-06,
      "loss": 0.7134,
      "step": 3630
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9061442906166154e-06,
      "loss": 0.6633,
      "step": 3635
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9058714954717345e-06,
      "loss": 0.6869,
      "step": 3640
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9055983240357123e-06,
      "loss": 0.6967,
      "step": 3645
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9053247764220218e-06,
      "loss": 0.7272,
      "step": 3650
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.905050852744292e-06,
      "loss": 0.6696,
      "step": 3655
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.904776553116309e-06,
      "loss": 0.6606,
      "step": 3660
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9045018776520138e-06,
      "loss": 0.7055,
      "step": 3665
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9042268264655048e-06,
      "loss": 0.6879,
      "step": 3670
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9039513996710357e-06,
      "loss": 0.7322,
      "step": 3675
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.903675597383016e-06,
      "loss": 0.7115,
      "step": 3680
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9033994197160124e-06,
      "loss": 0.6892,
      "step": 3685
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.903122866784746e-06,
      "loss": 0.7278,
      "step": 3690
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9028459387040944e-06,
      "loss": 0.6844,
      "step": 3695
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9025686355890916e-06,
      "loss": 0.7143,
      "step": 3700
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9022909575549265e-06,
      "loss": 0.7161,
      "step": 3705
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9020129047169443e-06,
      "loss": 0.7013,
      "step": 3710
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9017344771906463e-06,
      "loss": 0.6588,
      "step": 3715
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9014556750916879e-06,
      "loss": 0.6451,
      "step": 3720
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9011764985358817e-06,
      "loss": 0.6991,
      "step": 3725
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.900896947639195e-06,
      "loss": 0.6901,
      "step": 3730
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9006170225177508e-06,
      "loss": 0.7236,
      "step": 3735
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9003367232878273e-06,
      "loss": 0.6827,
      "step": 3740
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9000560500658592e-06,
      "loss": 0.6791,
      "step": 3745
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.8997750029684347e-06,
      "loss": 0.6822,
      "step": 3750
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.899493582112299e-06,
      "loss": 0.7139,
      "step": 3755
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.8992117876143516e-06,
      "loss": 0.6901,
      "step": 3760
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.8989296195916476e-06,
      "loss": 0.7012,
      "step": 3765
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.8986470781613973e-06,
      "loss": 0.7305,
      "step": 3770
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.8983641634409656e-06,
      "loss": 0.6812,
      "step": 3775
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.8980808755478726e-06,
      "loss": 0.6781,
      "step": 3780
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.8977972145997945e-06,
      "loss": 0.7271,
      "step": 3785
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.897513180714561e-06,
      "loss": 0.7162,
      "step": 3790
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.8972287740101572e-06,
      "loss": 0.6669,
      "step": 3795
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.8969439946047232e-06,
      "loss": 0.7358,
      "step": 3800
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.8966588426165544e-06,
      "loss": 0.7315,
      "step": 3805
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.8963733181640999e-06,
      "loss": 0.7245,
      "step": 3810
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.8960874213659643e-06,
      "loss": 0.7233,
      "step": 3815
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8958011523409067e-06,
      "loss": 0.7128,
      "step": 3820
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8955145112078408e-06,
      "loss": 0.6579,
      "step": 3825
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8952274980858344e-06,
      "loss": 0.7155,
      "step": 3830
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8949401130941109e-06,
      "loss": 0.6825,
      "step": 3835
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.894652356352047e-06,
      "loss": 0.6838,
      "step": 3840
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8943642279791747e-06,
      "loss": 0.7339,
      "step": 3845
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8940757280951799e-06,
      "loss": 0.7334,
      "step": 3850
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8937868568199026e-06,
      "loss": 0.7113,
      "step": 3855
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.893497614273338e-06,
      "loss": 0.7085,
      "step": 3860
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8932080005756346e-06,
      "loss": 0.7179,
      "step": 3865
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8929180158470953e-06,
      "loss": 0.6847,
      "step": 3870
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8926276602081777e-06,
      "loss": 0.7149,
      "step": 3875
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8923369337794926e-06,
      "loss": 0.6732,
      "step": 3880
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8920458366818055e-06,
      "loss": 0.6769,
      "step": 3885
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8917543690360351e-06,
      "loss": 0.6755,
      "step": 3890
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8914625309632552e-06,
      "loss": 0.7257,
      "step": 3895
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8911703225846921e-06,
      "loss": 0.6711,
      "step": 3900
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8908777440217274e-06,
      "loss": 0.6978,
      "step": 3905
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8905847953958951e-06,
      "loss": 0.706,
      "step": 3910
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8902914768288837e-06,
      "loss": 0.7315,
      "step": 3915
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8899977884425353e-06,
      "loss": 0.7385,
      "step": 3920
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8897037303588452e-06,
      "loss": 0.6868,
      "step": 3925
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.889409302699963e-06,
      "loss": 0.6924,
      "step": 3930
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8891145055881907e-06,
      "loss": 0.6575,
      "step": 3935
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8888193391459853e-06,
      "loss": 0.7152,
      "step": 3940
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8885238034959556e-06,
      "loss": 0.6853,
      "step": 3945
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8882278987608653e-06,
      "loss": 0.6626,
      "step": 3950
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8879316250636302e-06,
      "loss": 0.6714,
      "step": 3955
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8876349825273197e-06,
      "loss": 0.6994,
      "step": 3960
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8873379712751567e-06,
      "loss": 0.7311,
      "step": 3965
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8870405914305173e-06,
      "loss": 0.7214,
      "step": 3970
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.88674284311693e-06,
      "loss": 0.7004,
      "step": 3975
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8864447264580776e-06,
      "loss": 0.6956,
      "step": 3980
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8861462415777942e-06,
      "loss": 0.7378,
      "step": 3985
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.885847388600069e-06,
      "loss": 0.7195,
      "step": 3990
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8855481676490417e-06,
      "loss": 0.6948,
      "step": 3995
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.885248578849007e-06,
      "loss": 0.7367,
      "step": 4000
    },
    {
      "epoch": 0.16,
      "eval_loss": 0.666739821434021,
      "eval_runtime": 138.8501,
      "eval_samples_per_second": 17.04,
      "eval_steps_per_second": 2.845,
      "step": 4000
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.884948622324411e-06,
      "loss": 0.6942,
      "step": 4005
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.884648298199853e-06,
      "loss": 0.6911,
      "step": 4010
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8843476066000856e-06,
      "loss": 0.7338,
      "step": 4015
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.884046547650013e-06,
      "loss": 0.6914,
      "step": 4020
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8837451214746922e-06,
      "loss": 0.6998,
      "step": 4025
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8834433281993336e-06,
      "loss": 0.7024,
      "step": 4030
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8831411679492992e-06,
      "loss": 0.6748,
      "step": 4035
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.882838640850104e-06,
      "loss": 0.687,
      "step": 4040
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8825357470274148e-06,
      "loss": 0.661,
      "step": 4045
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8822324866070512e-06,
      "loss": 0.7209,
      "step": 4050
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8819288597149846e-06,
      "loss": 0.6597,
      "step": 4055
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.88162486647734e-06,
      "loss": 0.6923,
      "step": 4060
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8813205070203924e-06,
      "loss": 0.6875,
      "step": 4065
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8810157814705705e-06,
      "loss": 0.7085,
      "step": 4070
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8807106899544547e-06,
      "loss": 0.6786,
      "step": 4075
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8804052325987775e-06,
      "loss": 0.7078,
      "step": 4080
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8800994095304227e-06,
      "loss": 0.6945,
      "step": 4085
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8797932208764273e-06,
      "loss": 0.7232,
      "step": 4090
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8794866667639791e-06,
      "loss": 0.6876,
      "step": 4095
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8791797473204176e-06,
      "loss": 0.7299,
      "step": 4100
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8788724626732347e-06,
      "loss": 0.6689,
      "step": 4105
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.878564812950074e-06,
      "loss": 0.7345,
      "step": 4110
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8782567982787302e-06,
      "loss": 0.7271,
      "step": 4115
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8779484187871504e-06,
      "loss": 0.7328,
      "step": 4120
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8776396746034324e-06,
      "loss": 0.7043,
      "step": 4125
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8773305658558258e-06,
      "loss": 0.6841,
      "step": 4130
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8770210926727316e-06,
      "loss": 0.7154,
      "step": 4135
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8767112551827027e-06,
      "loss": 0.7329,
      "step": 4140
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8764010535144426e-06,
      "loss": 0.7226,
      "step": 4145
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8760904877968065e-06,
      "loss": 0.6958,
      "step": 4150
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8757795581588005e-06,
      "loss": 0.7373,
      "step": 4155
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8754682647295822e-06,
      "loss": 0.7313,
      "step": 4160
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.87515660763846e-06,
      "loss": 0.7333,
      "step": 4165
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8748445870148941e-06,
      "loss": 0.6882,
      "step": 4170
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8745322029884946e-06,
      "loss": 0.681,
      "step": 4175
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8742194556890233e-06,
      "loss": 0.7227,
      "step": 4180
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8739063452463926e-06,
      "loss": 0.7047,
      "step": 4185
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.873592871790666e-06,
      "loss": 0.6831,
      "step": 4190
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8732790354520579e-06,
      "loss": 0.6887,
      "step": 4195
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8729648363609324e-06,
      "loss": 0.6833,
      "step": 4200
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8726502746478058e-06,
      "loss": 0.6928,
      "step": 4205
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.872335350443344e-06,
      "loss": 0.7192,
      "step": 4210
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.872020063878364e-06,
      "loss": 0.6792,
      "step": 4215
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8717044150838326e-06,
      "loss": 0.6942,
      "step": 4220
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.871388404190868e-06,
      "loss": 0.733,
      "step": 4225
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8710720313307382e-06,
      "loss": 0.6587,
      "step": 4230
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8707552966348618e-06,
      "loss": 0.7068,
      "step": 4235
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.870438200234808e-06,
      "loss": 0.7121,
      "step": 4240
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8701207422622951e-06,
      "loss": 0.6572,
      "step": 4245
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.869802922849193e-06,
      "loss": 0.6988,
      "step": 4250
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8694847421275206e-06,
      "loss": 0.7009,
      "step": 4255
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8691662002294481e-06,
      "loss": 0.7128,
      "step": 4260
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8688472972872947e-06,
      "loss": 0.6869,
      "step": 4265
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8685280334335296e-06,
      "loss": 0.6775,
      "step": 4270
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8682084088007728e-06,
      "loss": 0.6787,
      "step": 4275
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8678884235217928e-06,
      "loss": 0.7285,
      "step": 4280
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8675680777295097e-06,
      "loss": 0.6764,
      "step": 4285
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8672473715569916e-06,
      "loss": 0.7242,
      "step": 4290
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8669263051374572e-06,
      "loss": 0.6771,
      "step": 4295
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8666048786042752e-06,
      "loss": 0.7394,
      "step": 4300
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.866283092090963e-06,
      "loss": 0.6957,
      "step": 4305
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8659609457311875e-06,
      "loss": 0.6816,
      "step": 4310
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.865638439658766e-06,
      "loss": 0.6952,
      "step": 4315
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8653155740076647e-06,
      "loss": 0.6754,
      "step": 4320
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8649923489119992e-06,
      "loss": 0.7266,
      "step": 4325
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8646687645060343e-06,
      "loss": 0.7218,
      "step": 4330
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8643448209241841e-06,
      "loss": 0.6779,
      "step": 4335
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.864020518301012e-06,
      "loss": 0.7003,
      "step": 4340
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8636958567712303e-06,
      "loss": 0.7015,
      "step": 4345
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8633708364697013e-06,
      "loss": 0.6808,
      "step": 4350
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8630454575314344e-06,
      "loss": 0.6816,
      "step": 4355
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8627197200915902e-06,
      "loss": 0.7126,
      "step": 4360
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.862393624285477e-06,
      "loss": 0.6838,
      "step": 4365
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8620671702485517e-06,
      "loss": 0.6578,
      "step": 4370
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.861740358116421e-06,
      "loss": 0.7344,
      "step": 4375
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8614131880248393e-06,
      "loss": 0.7172,
      "step": 4380
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8610856601097108e-06,
      "loss": 0.7621,
      "step": 4385
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8607577745070873e-06,
      "loss": 0.7071,
      "step": 4390
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8604295313531698e-06,
      "loss": 0.6889,
      "step": 4395
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8601009307843078e-06,
      "loss": 0.6992,
      "step": 4400
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8597719729369988e-06,
      "loss": 0.6843,
      "step": 4405
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8594426579478891e-06,
      "loss": 0.6791,
      "step": 4410
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8591129859537738e-06,
      "loss": 0.7045,
      "step": 4415
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.858782957091595e-06,
      "loss": 0.6975,
      "step": 4420
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.858452571498444e-06,
      "loss": 0.7386,
      "step": 4425
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8581218293115607e-06,
      "loss": 0.6974,
      "step": 4430
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8577907306683317e-06,
      "loss": 0.7313,
      "step": 4435
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.857459275706293e-06,
      "loss": 0.6941,
      "step": 4440
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8571274645631281e-06,
      "loss": 0.7037,
      "step": 4445
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8567952973766685e-06,
      "loss": 0.655,
      "step": 4450
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.856462774284893e-06,
      "loss": 0.712,
      "step": 4455
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8561298954259297e-06,
      "loss": 0.68,
      "step": 4460
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8557966609380528e-06,
      "loss": 0.6917,
      "step": 4465
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8554630709596855e-06,
      "loss": 0.6719,
      "step": 4470
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8551291256293977e-06,
      "loss": 0.7117,
      "step": 4475
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.854794825085908e-06,
      "loss": 0.7157,
      "step": 4480
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8544601694680814e-06,
      "loss": 0.7147,
      "step": 4485
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8541251589149313e-06,
      "loss": 0.6964,
      "step": 4490
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.853789793565618e-06,
      "loss": 0.7083,
      "step": 4495
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8534540735594492e-06,
      "loss": 0.6735,
      "step": 4500
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8531179990358806e-06,
      "loss": 0.6719,
      "step": 4505
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.852781570134514e-06,
      "loss": 0.6729,
      "step": 4510
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8524447869950995e-06,
      "loss": 0.6988,
      "step": 4515
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8521076497575335e-06,
      "loss": 0.6907,
      "step": 4520
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8517701585618602e-06,
      "loss": 0.7268,
      "step": 4525
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8514323135482704e-06,
      "loss": 0.7261,
      "step": 4530
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8510941148571018e-06,
      "loss": 0.6763,
      "step": 4535
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8507555626288397e-06,
      "loss": 0.6801,
      "step": 4540
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8504166570041152e-06,
      "loss": 0.7502,
      "step": 4545
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8500773981237069e-06,
      "loss": 0.6682,
      "step": 4550
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8497377861285401e-06,
      "loss": 0.709,
      "step": 4555
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8493978211596865e-06,
      "loss": 0.7001,
      "step": 4560
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.849057503358365e-06,
      "loss": 0.7111,
      "step": 4565
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8487168328659403e-06,
      "loss": 0.701,
      "step": 4570
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8483758098239237e-06,
      "loss": 0.691,
      "step": 4575
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8480344343739738e-06,
      "loss": 0.6979,
      "step": 4580
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8476927066578946e-06,
      "loss": 0.7074,
      "step": 4585
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8473506268176372e-06,
      "loss": 0.7229,
      "step": 4590
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8470081949952982e-06,
      "loss": 0.7015,
      "step": 4595
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.846665411333121e-06,
      "loss": 0.7329,
      "step": 4600
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8463222759734948e-06,
      "loss": 0.6908,
      "step": 4605
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8459787890589554e-06,
      "loss": 0.7381,
      "step": 4610
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8456349507321836e-06,
      "loss": 0.6663,
      "step": 4615
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8452907611360076e-06,
      "loss": 0.7021,
      "step": 4620
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8449462204134002e-06,
      "loss": 0.6826,
      "step": 4625
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8446013287074811e-06,
      "loss": 0.6959,
      "step": 4630
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8442560861615148e-06,
      "loss": 0.7348,
      "step": 4635
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8439104929189124e-06,
      "loss": 0.6699,
      "step": 4640
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8435645491232307e-06,
      "loss": 0.7354,
      "step": 4645
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8432182549181707e-06,
      "loss": 0.6721,
      "step": 4650
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8428716104475806e-06,
      "loss": 0.7104,
      "step": 4655
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8425246158554537e-06,
      "loss": 0.6808,
      "step": 4660
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8421772712859282e-06,
      "loss": 0.6941,
      "step": 4665
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8418295768832883e-06,
      "loss": 0.687,
      "step": 4670
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.841481532791963e-06,
      "loss": 0.6894,
      "step": 4675
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.841133139156527e-06,
      "loss": 0.6663,
      "step": 4680
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8407843961216995e-06,
      "loss": 0.6423,
      "step": 4685
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8404353038323459e-06,
      "loss": 0.7073,
      "step": 4690
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8400858624334758e-06,
      "loss": 0.7149,
      "step": 4695
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8397360720702442e-06,
      "loss": 0.7024,
      "step": 4700
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8393859328879511e-06,
      "loss": 0.7039,
      "step": 4705
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.839035445032041e-06,
      "loss": 0.6758,
      "step": 4710
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8386846086481036e-06,
      "loss": 0.6833,
      "step": 4715
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8383334238818736e-06,
      "loss": 0.6996,
      "step": 4720
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8379818908792295e-06,
      "loss": 0.692,
      "step": 4725
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8376300097861953e-06,
      "loss": 0.7099,
      "step": 4730
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8372777807489396e-06,
      "loss": 0.6859,
      "step": 4735
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8369252039137753e-06,
      "loss": 0.696,
      "step": 4740
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8365722794271594e-06,
      "loss": 0.6713,
      "step": 4745
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8362190074356935e-06,
      "loss": 0.7215,
      "step": 4750
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8358653880861245e-06,
      "loss": 0.7001,
      "step": 4755
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.835511421525342e-06,
      "loss": 0.727,
      "step": 4760
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8351571079003812e-06,
      "loss": 0.6764,
      "step": 4765
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8348024473584208e-06,
      "loss": 0.6667,
      "step": 4770
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.834447440046783e-06,
      "loss": 0.7291,
      "step": 4775
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8340920861129358e-06,
      "loss": 0.6699,
      "step": 4780
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8337363857044894e-06,
      "loss": 0.6825,
      "step": 4785
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.833380338969199e-06,
      "loss": 0.6954,
      "step": 4790
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.833023946054963e-06,
      "loss": 0.7002,
      "step": 4795
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8326672071098246e-06,
      "loss": 0.699,
      "step": 4800
    },
    {
      "epoch": 0.19,
      "eval_loss": 0.6643534302711487,
      "eval_runtime": 138.4591,
      "eval_samples_per_second": 17.088,
      "eval_steps_per_second": 2.853,
      "step": 4800
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8323101222819693e-06,
      "loss": 0.677,
      "step": 4805
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8319526917197275e-06,
      "loss": 0.7381,
      "step": 4810
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8315949155715722e-06,
      "loss": 0.6681,
      "step": 4815
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8312367939861214e-06,
      "loss": 0.6696,
      "step": 4820
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8308783271121346e-06,
      "loss": 0.713,
      "step": 4825
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.830519515098517e-06,
      "loss": 0.7155,
      "step": 4830
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8301603580943148e-06,
      "loss": 0.694,
      "step": 4835
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8298008562487195e-06,
      "loss": 0.7343,
      "step": 4840
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.829441009711065e-06,
      "loss": 0.7173,
      "step": 4845
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8290808186308276e-06,
      "loss": 0.7167,
      "step": 4850
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.828720283157629e-06,
      "loss": 0.7265,
      "step": 4855
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8283594034412313e-06,
      "loss": 0.6793,
      "step": 4860
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8279981796315412e-06,
      "loss": 0.6807,
      "step": 4865
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8276366118786078e-06,
      "loss": 0.6737,
      "step": 4870
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8272747003326235e-06,
      "loss": 0.7202,
      "step": 4875
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8269124451439231e-06,
      "loss": 0.7226,
      "step": 4880
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8265498464629837e-06,
      "loss": 0.6772,
      "step": 4885
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8261869044404265e-06,
      "loss": 0.6868,
      "step": 4890
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.825823619227014e-06,
      "loss": 0.703,
      "step": 4895
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.825459990973652e-06,
      "loss": 0.7193,
      "step": 4900
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8250960198313878e-06,
      "loss": 0.7194,
      "step": 4905
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8247317059514126e-06,
      "loss": 0.6859,
      "step": 4910
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8243670494850592e-06,
      "loss": 0.6749,
      "step": 4915
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.824002050583802e-06,
      "loss": 0.6542,
      "step": 4920
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8236367093992592e-06,
      "loss": 0.6592,
      "step": 4925
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.82327102608319e-06,
      "loss": 0.6788,
      "step": 4930
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8229050007874961e-06,
      "loss": 0.6909,
      "step": 4935
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.822538633664221e-06,
      "loss": 0.7105,
      "step": 4940
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8221719248655507e-06,
      "loss": 0.6731,
      "step": 4945
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.821804874543813e-06,
      "loss": 0.685,
      "step": 4950
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.821437482851477e-06,
      "loss": 0.7251,
      "step": 4955
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.821069749941154e-06,
      "loss": 0.6517,
      "step": 4960
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8207016759655973e-06,
      "loss": 0.704,
      "step": 4965
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8203332610777017e-06,
      "loss": 0.7063,
      "step": 4970
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8199645054305037e-06,
      "loss": 0.7208,
      "step": 4975
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8195954091771805e-06,
      "loss": 0.7339,
      "step": 4980
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8192259724710518e-06,
      "loss": 0.6627,
      "step": 4985
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.818856195465579e-06,
      "loss": 0.7021,
      "step": 4990
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8184860783143635e-06,
      "loss": 0.7108,
      "step": 4995
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8181156211711488e-06,
      "loss": 0.639,
      "step": 5000
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8177448241898196e-06,
      "loss": 0.7256,
      "step": 5005
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.817373687524402e-06,
      "loss": 0.6977,
      "step": 5010
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.817002211329063e-06,
      "loss": 0.6984,
      "step": 5015
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8166303957581103e-06,
      "loss": 0.6875,
      "step": 5020
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8162582409659932e-06,
      "loss": 0.6602,
      "step": 5025
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.815885747107301e-06,
      "loss": 0.6819,
      "step": 5030
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.815512914336765e-06,
      "loss": 0.6843,
      "step": 5035
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8151397428092563e-06,
      "loss": 0.7079,
      "step": 5040
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8147662326797872e-06,
      "loss": 0.7022,
      "step": 5045
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8143923841035107e-06,
      "loss": 0.6897,
      "step": 5050
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8140181972357201e-06,
      "loss": 0.6774,
      "step": 5055
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8136436722318496e-06,
      "loss": 0.6877,
      "step": 5060
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.813268809247473e-06,
      "loss": 0.6957,
      "step": 5065
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8128936084383058e-06,
      "loss": 0.7274,
      "step": 5070
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.812518069960203e-06,
      "loss": 0.6767,
      "step": 5075
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8121421939691599e-06,
      "loss": 0.6857,
      "step": 5080
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8117659806213122e-06,
      "loss": 0.6637,
      "step": 5085
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8113894300729356e-06,
      "loss": 0.6777,
      "step": 5090
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8110125424804458e-06,
      "loss": 0.6666,
      "step": 5095
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8106353180003988e-06,
      "loss": 0.6811,
      "step": 5100
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8102577567894905e-06,
      "loss": 0.6735,
      "step": 5105
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8098798590045563e-06,
      "loss": 0.7543,
      "step": 5110
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8095016248025717e-06,
      "loss": 0.6957,
      "step": 5115
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.809123054340652e-06,
      "loss": 0.6969,
      "step": 5120
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8087441477760517e-06,
      "loss": 0.6775,
      "step": 5125
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8083649052661661e-06,
      "loss": 0.6794,
      "step": 5130
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8079853269685282e-06,
      "loss": 0.6253,
      "step": 5135
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8076054130408123e-06,
      "loss": 0.7048,
      "step": 5140
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8072251636408314e-06,
      "loss": 0.681,
      "step": 5145
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.806844578926537e-06,
      "loss": 0.6644,
      "step": 5150
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8064636590560215e-06,
      "loss": 0.6972,
      "step": 5155
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8060824041875154e-06,
      "loss": 0.6831,
      "step": 5160
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8057008144793882e-06,
      "loss": 0.6738,
      "step": 5165
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8053188900901497e-06,
      "loss": 0.7293,
      "step": 5170
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8049366311784477e-06,
      "loss": 0.6975,
      "step": 5175
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8045540379030691e-06,
      "loss": 0.7275,
      "step": 5180
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8041711104229401e-06,
      "loss": 0.6867,
      "step": 5185
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.803787848897125e-06,
      "loss": 0.6857,
      "step": 5190
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8034042534848277e-06,
      "loss": 0.6861,
      "step": 5195
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8030203243453901e-06,
      "loss": 0.6732,
      "step": 5200
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8026360616382933e-06,
      "loss": 0.7133,
      "step": 5205
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8022514655231565e-06,
      "loss": 0.7196,
      "step": 5210
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8018665361597373e-06,
      "loss": 0.6929,
      "step": 5215
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.801481273707933e-06,
      "loss": 0.7227,
      "step": 5220
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8010956783277772e-06,
      "loss": 0.6655,
      "step": 5225
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8007097501794435e-06,
      "loss": 0.7454,
      "step": 5230
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8003234894232426e-06,
      "loss": 0.6815,
      "step": 5235
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.7999368962196243e-06,
      "loss": 0.7456,
      "step": 5240
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.7995499707291762e-06,
      "loss": 0.6822,
      "step": 5245
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.799162713112623e-06,
      "loss": 0.7083,
      "step": 5250
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.798775123530829e-06,
      "loss": 0.6874,
      "step": 5255
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.798387202144795e-06,
      "loss": 0.6864,
      "step": 5260
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.7979989491156603e-06,
      "loss": 0.6991,
      "step": 5265
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.797610364604702e-06,
      "loss": 0.7077,
      "step": 5270
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.7972214487733345e-06,
      "loss": 0.6461,
      "step": 5275
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.7968322017831102e-06,
      "loss": 0.6968,
      "step": 5280
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.7964426237957188e-06,
      "loss": 0.6635,
      "step": 5285
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.7960527149729878e-06,
      "loss": 0.684,
      "step": 5290
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7956624754768818e-06,
      "loss": 0.6784,
      "step": 5295
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.795271905469503e-06,
      "loss": 0.7481,
      "step": 5300
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7948810051130905e-06,
      "loss": 0.683,
      "step": 5305
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.794489774570021e-06,
      "loss": 0.6661,
      "step": 5310
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7940982140028087e-06,
      "loss": 0.7184,
      "step": 5315
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.793706323574104e-06,
      "loss": 0.6612,
      "step": 5320
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7933141034466948e-06,
      "loss": 0.703,
      "step": 5325
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.792921553783506e-06,
      "loss": 0.7108,
      "step": 5330
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7925286747475994e-06,
      "loss": 0.7301,
      "step": 5335
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7921354665021735e-06,
      "loss": 0.6704,
      "step": 5340
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7917419292105636e-06,
      "loss": 0.7065,
      "step": 5345
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7913480630362417e-06,
      "loss": 0.7033,
      "step": 5350
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.790953868142816e-06,
      "loss": 0.7258,
      "step": 5355
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.790559344694032e-06,
      "loss": 0.6887,
      "step": 5360
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7901644928537715e-06,
      "loss": 0.7195,
      "step": 5365
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7897693127860524e-06,
      "loss": 0.7058,
      "step": 5370
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7893738046550286e-06,
      "loss": 0.7177,
      "step": 5375
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7889779686249912e-06,
      "loss": 0.6891,
      "step": 5380
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.788581804860367e-06,
      "loss": 0.6551,
      "step": 5385
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7881853135257185e-06,
      "loss": 0.7061,
      "step": 5390
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7877884947857455e-06,
      "loss": 0.6706,
      "step": 5395
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7873913488052827e-06,
      "loss": 0.6719,
      "step": 5400
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7869938757493011e-06,
      "loss": 0.7022,
      "step": 5405
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7865960757829075e-06,
      "loss": 0.676,
      "step": 5410
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7861979490713445e-06,
      "loss": 0.7427,
      "step": 5415
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7857994957799906e-06,
      "loss": 0.698,
      "step": 5420
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.78540071607436e-06,
      "loss": 0.7023,
      "step": 5425
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.785001610120102e-06,
      "loss": 0.6985,
      "step": 5430
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.784602178083002e-06,
      "loss": 0.7192,
      "step": 5435
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7842024201289801e-06,
      "loss": 0.6682,
      "step": 5440
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7838023364240929e-06,
      "loss": 0.694,
      "step": 5445
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7834019271345313e-06,
      "loss": 0.6484,
      "step": 5450
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7830011924266221e-06,
      "loss": 0.6921,
      "step": 5455
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7826001324668267e-06,
      "loss": 0.7135,
      "step": 5460
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7821987474217424e-06,
      "loss": 0.6631,
      "step": 5465
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7817970374581006e-06,
      "loss": 0.6572,
      "step": 5470
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.781395002742768e-06,
      "loss": 0.7068,
      "step": 5475
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.780992643442747e-06,
      "loss": 0.6417,
      "step": 5480
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7805899597251735e-06,
      "loss": 0.6847,
      "step": 5485
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7801869517573191e-06,
      "loss": 0.6862,
      "step": 5490
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.77978361970659e-06,
      "loss": 0.6696,
      "step": 5495
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7793799637405265e-06,
      "loss": 0.737,
      "step": 5500
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7789759840268037e-06,
      "loss": 0.6826,
      "step": 5505
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7785716807332315e-06,
      "loss": 0.7288,
      "step": 5510
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.778167054027754e-06,
      "loss": 0.6856,
      "step": 5515
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7777621040784496e-06,
      "loss": 0.6801,
      "step": 5520
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.777356831053531e-06,
      "loss": 0.6737,
      "step": 5525
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7769512351213448e-06,
      "loss": 0.6547,
      "step": 5530
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7765453164503722e-06,
      "loss": 0.7255,
      "step": 5535
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7761390752092284e-06,
      "loss": 0.6757,
      "step": 5540
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7757325115666624e-06,
      "loss": 0.6752,
      "step": 5545
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7753256256915572e-06,
      "loss": 0.6901,
      "step": 5550
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7749184177529294e-06,
      "loss": 0.6747,
      "step": 5555
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7745108879199302e-06,
      "loss": 0.6852,
      "step": 5560
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7741030363618437e-06,
      "loss": 0.6804,
      "step": 5565
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7736948632480874e-06,
      "loss": 0.6783,
      "step": 5570
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7732863687482138e-06,
      "loss": 0.6799,
      "step": 5575
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.772877553031907e-06,
      "loss": 0.7125,
      "step": 5580
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7724684162689863e-06,
      "loss": 0.6851,
      "step": 5585
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.772058958629403e-06,
      "loss": 0.6906,
      "step": 5590
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7716491802832425e-06,
      "loss": 0.7292,
      "step": 5595
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.771239081400723e-06,
      "loss": 0.7028,
      "step": 5600
    },
    {
      "epoch": 0.23,
      "eval_loss": 0.6604963541030884,
      "eval_runtime": 138.4769,
      "eval_samples_per_second": 17.086,
      "eval_steps_per_second": 2.852,
      "step": 5600
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7708286621521964e-06,
      "loss": 0.6984,
      "step": 5605
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7704179227081467e-06,
      "loss": 0.7075,
      "step": 5610
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7700068632391917e-06,
      "loss": 0.7014,
      "step": 5615
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7695954839160824e-06,
      "loss": 0.6986,
      "step": 5620
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7691837849097015e-06,
      "loss": 0.7069,
      "step": 5625
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7687717663910658e-06,
      "loss": 0.6893,
      "step": 5630
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7683594285313235e-06,
      "loss": 0.6479,
      "step": 5635
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7679467715017567e-06,
      "loss": 0.7326,
      "step": 5640
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7675337954737795e-06,
      "loss": 0.7032,
      "step": 5645
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7671205006189384e-06,
      "loss": 0.6587,
      "step": 5650
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7667068871089126e-06,
      "loss": 0.6494,
      "step": 5655
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7662929551155136e-06,
      "loss": 0.7001,
      "step": 5660
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7658787048106846e-06,
      "loss": 0.7043,
      "step": 5665
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7654641363665027e-06,
      "loss": 0.6735,
      "step": 5670
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.765049249955175e-06,
      "loss": 0.706,
      "step": 5675
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.764634045749042e-06,
      "loss": 0.7178,
      "step": 5680
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7642185239205763e-06,
      "loss": 0.6817,
      "step": 5685
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7638026846423813e-06,
      "loss": 0.739,
      "step": 5690
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7633865280871935e-06,
      "loss": 0.704,
      "step": 5695
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.762970054427881e-06,
      "loss": 0.6859,
      "step": 5700
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7625532638374429e-06,
      "loss": 0.6645,
      "step": 5705
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7621361564890104e-06,
      "loss": 0.6885,
      "step": 5710
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7617187325558463e-06,
      "loss": 0.7211,
      "step": 5715
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7613009922113449e-06,
      "loss": 0.6563,
      "step": 5720
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.760882935629032e-06,
      "loss": 0.6872,
      "step": 5725
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7604645629825647e-06,
      "loss": 0.7238,
      "step": 5730
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7600458744457312e-06,
      "loss": 0.6829,
      "step": 5735
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7596268701924513e-06,
      "loss": 0.6752,
      "step": 5740
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7592075503967751e-06,
      "loss": 0.7009,
      "step": 5745
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7587879152328852e-06,
      "loss": 0.6987,
      "step": 5750
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.758367964875094e-06,
      "loss": 0.7154,
      "step": 5755
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7579476994978454e-06,
      "loss": 0.6551,
      "step": 5760
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7575271192757138e-06,
      "loss": 0.6711,
      "step": 5765
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7571062243834046e-06,
      "loss": 0.7064,
      "step": 5770
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7566850149957536e-06,
      "loss": 0.6889,
      "step": 5775
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7562634912877282e-06,
      "loss": 0.6998,
      "step": 5780
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7558416534344252e-06,
      "loss": 0.6513,
      "step": 5785
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7554195016110725e-06,
      "loss": 0.7019,
      "step": 5790
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7549970359930285e-06,
      "loss": 0.7119,
      "step": 5795
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7545742567557811e-06,
      "loss": 0.6858,
      "step": 5800
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7541511640749499e-06,
      "loss": 0.6887,
      "step": 5805
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.753727758126283e-06,
      "loss": 0.6976,
      "step": 5810
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7533040390856608e-06,
      "loss": 0.6699,
      "step": 5815
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7528800071290914e-06,
      "loss": 0.6802,
      "step": 5820
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7524556624327147e-06,
      "loss": 0.6804,
      "step": 5825
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7520310051727992e-06,
      "loss": 0.6678,
      "step": 5830
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7516060355257443e-06,
      "loss": 0.6856,
      "step": 5835
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7511807536680782e-06,
      "loss": 0.6983,
      "step": 5840
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.75075515977646e-06,
      "loss": 0.7212,
      "step": 5845
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7503292540276772e-06,
      "loss": 0.6783,
      "step": 5850
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7499030365986475e-06,
      "loss": 0.6932,
      "step": 5855
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.749476507666418e-06,
      "loss": 0.7191,
      "step": 5860
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7490496674081648e-06,
      "loss": 0.6817,
      "step": 5865
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.748622516001194e-06,
      "loss": 0.6568,
      "step": 5870
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7481950536229402e-06,
      "loss": 0.6861,
      "step": 5875
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7477672804509679e-06,
      "loss": 0.6995,
      "step": 5880
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7473391966629698e-06,
      "loss": 0.72,
      "step": 5885
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7469108024367687e-06,
      "loss": 0.6983,
      "step": 5890
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.746482097950316e-06,
      "loss": 0.6961,
      "step": 5895
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7460530833816915e-06,
      "loss": 0.7183,
      "step": 5900
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.745623758909104e-06,
      "loss": 0.7194,
      "step": 5905
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7451941247108914e-06,
      "loss": 0.6699,
      "step": 5910
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.74476418096552e-06,
      "loss": 0.6807,
      "step": 5915
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7443339278515846e-06,
      "loss": 0.6835,
      "step": 5920
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7439033655478084e-06,
      "loss": 0.6874,
      "step": 5925
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7434724942330436e-06,
      "loss": 0.6832,
      "step": 5930
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7430413140862703e-06,
      "loss": 0.6661,
      "step": 5935
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7426098252865968e-06,
      "loss": 0.677,
      "step": 5940
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.74217802801326e-06,
      "loss": 0.6587,
      "step": 5945
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7417459224456246e-06,
      "loss": 0.6613,
      "step": 5950
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.741313508763184e-06,
      "loss": 0.7241,
      "step": 5955
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.740880787145558e-06,
      "loss": 0.7194,
      "step": 5960
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7404477577724964e-06,
      "loss": 0.6832,
      "step": 5965
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7400144208238751e-06,
      "loss": 0.6761,
      "step": 5970
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7395807764796993e-06,
      "loss": 0.7259,
      "step": 5975
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7391468249201004e-06,
      "loss": 0.6704,
      "step": 5980
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7387125663253386e-06,
      "loss": 0.6836,
      "step": 5985
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7382780008758005e-06,
      "loss": 0.6555,
      "step": 5990
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7378431287520016e-06,
      "loss": 0.7281,
      "step": 5995
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7374079501345835e-06,
      "loss": 0.6644,
      "step": 6000
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7369724652043156e-06,
      "loss": 0.6314,
      "step": 6005
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7365366741420947e-06,
      "loss": 0.6719,
      "step": 6010
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.736100577128945e-06,
      "loss": 0.6913,
      "step": 6015
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7356641743460166e-06,
      "loss": 0.6858,
      "step": 6020
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7352274659745878e-06,
      "loss": 0.6853,
      "step": 6025
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7347904521960635e-06,
      "loss": 0.6921,
      "step": 6030
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7343531331919756e-06,
      "loss": 0.6898,
      "step": 6035
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7339155091439823e-06,
      "loss": 0.6996,
      "step": 6040
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.733477580233869e-06,
      "loss": 0.7071,
      "step": 6045
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7330393466435474e-06,
      "loss": 0.6903,
      "step": 6050
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7326008085550564e-06,
      "loss": 0.6929,
      "step": 6055
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.73216196615056e-06,
      "loss": 0.7195,
      "step": 6060
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7317228196123504e-06,
      "loss": 0.6737,
      "step": 6065
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7312833691228445e-06,
      "loss": 0.7118,
      "step": 6070
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7308436148645871e-06,
      "loss": 0.7004,
      "step": 6075
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7304035570202476e-06,
      "loss": 0.666,
      "step": 6080
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7299631957726223e-06,
      "loss": 0.6574,
      "step": 6085
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7295225313046337e-06,
      "loss": 0.7057,
      "step": 6090
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.72908156379933e-06,
      "loss": 0.6804,
      "step": 6095
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7286402934398848e-06,
      "loss": 0.7117,
      "step": 6100
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7281987204095985e-06,
      "loss": 0.7004,
      "step": 6105
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7277568448918962e-06,
      "loss": 0.6945,
      "step": 6110
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7273146670703295e-06,
      "loss": 0.6268,
      "step": 6115
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7268721871285753e-06,
      "loss": 0.6772,
      "step": 6120
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7264294052504358e-06,
      "loss": 0.7141,
      "step": 6125
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7259863216198385e-06,
      "loss": 0.6466,
      "step": 6130
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7255429364208366e-06,
      "loss": 0.6835,
      "step": 6135
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7250992498376086e-06,
      "loss": 0.6938,
      "step": 6140
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7246552620544584e-06,
      "loss": 0.7017,
      "step": 6145
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7242109732558137e-06,
      "loss": 0.7161,
      "step": 6150
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7237663836262289e-06,
      "loss": 0.6646,
      "step": 6155
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7233214933503826e-06,
      "loss": 0.6712,
      "step": 6160
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7228763026130783e-06,
      "loss": 0.6749,
      "step": 6165
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7224308115992443e-06,
      "loss": 0.647,
      "step": 6170
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7219850204939338e-06,
      "loss": 0.7009,
      "step": 6175
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7215389294823243e-06,
      "loss": 0.6568,
      "step": 6180
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7210925387497186e-06,
      "loss": 0.6501,
      "step": 6185
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7206458484815429e-06,
      "loss": 0.6926,
      "step": 6190
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7201988588633489e-06,
      "loss": 0.6973,
      "step": 6195
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7197515700808124e-06,
      "loss": 0.6979,
      "step": 6200
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7193039823197323e-06,
      "loss": 0.6634,
      "step": 6205
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7188560957660338e-06,
      "loss": 0.6958,
      "step": 6210
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7184079106057645e-06,
      "loss": 0.7073,
      "step": 6215
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7179594270250965e-06,
      "loss": 0.7147,
      "step": 6220
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7175106452103268e-06,
      "loss": 0.7319,
      "step": 6225
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7170615653478742e-06,
      "loss": 0.6786,
      "step": 6230
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7166121876242837e-06,
      "loss": 0.6643,
      "step": 6235
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.716162512226222e-06,
      "loss": 0.7524,
      "step": 6240
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.715712539340481e-06,
      "loss": 0.6517,
      "step": 6245
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.715262269153975e-06,
      "loss": 0.6999,
      "step": 6250
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7148117018537432e-06,
      "loss": 0.6756,
      "step": 6255
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7143608376269462e-06,
      "loss": 0.7208,
      "step": 6260
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.71390967666087e-06,
      "loss": 0.6793,
      "step": 6265
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7134582191429223e-06,
      "loss": 0.68,
      "step": 6270
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7130064652606352e-06,
      "loss": 0.6666,
      "step": 6275
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7125544152016627e-06,
      "loss": 0.6714,
      "step": 6280
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7121020691537831e-06,
      "loss": 0.7034,
      "step": 6285
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7116494273048966e-06,
      "loss": 0.6819,
      "step": 6290
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7111964898430266e-06,
      "loss": 0.66,
      "step": 6295
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.71074325695632e-06,
      "loss": 0.7156,
      "step": 6300
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7102897288330454e-06,
      "loss": 0.6804,
      "step": 6305
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7098359056615942e-06,
      "loss": 0.6807,
      "step": 6310
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7093817876304807e-06,
      "loss": 0.6701,
      "step": 6315
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7089273749283418e-06,
      "loss": 0.6785,
      "step": 6320
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7084726677439364e-06,
      "loss": 0.6941,
      "step": 6325
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7080176662661463e-06,
      "loss": 0.6811,
      "step": 6330
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7075623706839745e-06,
      "loss": 0.6797,
      "step": 6335
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7071067811865474e-06,
      "loss": 0.6913,
      "step": 6340
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7066508979631129e-06,
      "loss": 0.6917,
      "step": 6345
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7061947212030402e-06,
      "loss": 0.6709,
      "step": 6350
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.705738251095822e-06,
      "loss": 0.7023,
      "step": 6355
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7052814878310718e-06,
      "loss": 0.7109,
      "step": 6360
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7048244315985247e-06,
      "loss": 0.6795,
      "step": 6365
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7043670825880384e-06,
      "loss": 0.7013,
      "step": 6370
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7039094409895914e-06,
      "loss": 0.703,
      "step": 6375
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.703451506993284e-06,
      "loss": 0.732,
      "step": 6380
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7029932807893382e-06,
      "loss": 0.6905,
      "step": 6385
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.702534762568097e-06,
      "loss": 0.7172,
      "step": 6390
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7020759525200253e-06,
      "loss": 0.6674,
      "step": 6395
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.701616850835708e-06,
      "loss": 0.6694,
      "step": 6400
    },
    {
      "epoch": 0.26,
      "eval_loss": 0.6586260795593262,
      "eval_runtime": 139.2072,
      "eval_samples_per_second": 16.996,
      "eval_steps_per_second": 2.837,
      "step": 6400
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7011574577058525e-06,
      "loss": 0.7282,
      "step": 6405
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7006977733212867e-06,
      "loss": 0.6736,
      "step": 6410
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7002377978729596e-06,
      "loss": 0.6672,
      "step": 6415
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.6997775315519408e-06,
      "loss": 0.712,
      "step": 6420
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.6993169745494209e-06,
      "loss": 0.6949,
      "step": 6425
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.6988561270567115e-06,
      "loss": 0.6951,
      "step": 6430
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.698394989265244e-06,
      "loss": 0.7019,
      "step": 6435
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.6979335613665717e-06,
      "loss": 0.7143,
      "step": 6440
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.6974718435523678e-06,
      "loss": 0.7253,
      "step": 6445
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.6970098360144253e-06,
      "loss": 0.7103,
      "step": 6450
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.6965475389446586e-06,
      "loss": 0.6953,
      "step": 6455
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.6960849525351018e-06,
      "loss": 0.7247,
      "step": 6460
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.6956220769779088e-06,
      "loss": 0.7015,
      "step": 6465
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.6951589124653547e-06,
      "loss": 0.704,
      "step": 6470
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.6946954591898336e-06,
      "loss": 0.7054,
      "step": 6475
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.6942317173438604e-06,
      "loss": 0.6704,
      "step": 6480
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.693767687120069e-06,
      "loss": 0.6564,
      "step": 6485
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.6933033687112134e-06,
      "loss": 0.7054,
      "step": 6490
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.6928387623101681e-06,
      "loss": 0.676,
      "step": 6495
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.692373868109926e-06,
      "loss": 0.7034,
      "step": 6500
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.6919086863036003e-06,
      "loss": 0.6781,
      "step": 6505
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.6914432170844233e-06,
      "loss": 0.7057,
      "step": 6510
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.690977460645747e-06,
      "loss": 0.6929,
      "step": 6515
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.6905114171810429e-06,
      "loss": 0.6816,
      "step": 6520
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6900450868839009e-06,
      "loss": 0.6845,
      "step": 6525
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6895784699480306e-06,
      "loss": 0.6665,
      "step": 6530
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6891115665672608e-06,
      "loss": 0.6597,
      "step": 6535
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6886443769355393e-06,
      "loss": 0.6623,
      "step": 6540
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.688176901246932e-06,
      "loss": 0.6535,
      "step": 6545
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6877091396956247e-06,
      "loss": 0.6902,
      "step": 6550
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6872410924759215e-06,
      "loss": 0.6447,
      "step": 6555
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.686772759782245e-06,
      "loss": 0.7013,
      "step": 6560
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6863041418091366e-06,
      "loss": 0.7158,
      "step": 6565
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.685835238751256e-06,
      "loss": 0.7109,
      "step": 6570
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6853660508033816e-06,
      "loss": 0.7231,
      "step": 6575
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6848965781604099e-06,
      "loss": 0.7305,
      "step": 6580
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6844268210173556e-06,
      "loss": 0.7013,
      "step": 6585
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6839567795693524e-06,
      "loss": 0.6849,
      "step": 6590
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6834864540116506e-06,
      "loss": 0.7316,
      "step": 6595
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6830158445396196e-06,
      "loss": 0.7193,
      "step": 6600
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6825449513487466e-06,
      "loss": 0.7277,
      "step": 6605
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6820737746346368e-06,
      "loss": 0.7037,
      "step": 6610
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6816023145930123e-06,
      "loss": 0.7078,
      "step": 6615
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.681130571419714e-06,
      "loss": 0.6651,
      "step": 6620
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6806585453106997e-06,
      "loss": 0.6851,
      "step": 6625
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.680186236462045e-06,
      "loss": 0.6877,
      "step": 6630
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6797136450699427e-06,
      "loss": 0.6976,
      "step": 6635
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6792407713307036e-06,
      "loss": 0.7244,
      "step": 6640
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.678767615440755e-06,
      "loss": 0.6734,
      "step": 6645
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6782941775966416e-06,
      "loss": 0.6733,
      "step": 6650
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6778204579950255e-06,
      "loss": 0.6836,
      "step": 6655
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6773464568326859e-06,
      "loss": 0.7101,
      "step": 6660
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6768721743065186e-06,
      "loss": 0.695,
      "step": 6665
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6763976106135366e-06,
      "loss": 0.6774,
      "step": 6670
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6759227659508692e-06,
      "loss": 0.674,
      "step": 6675
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6754476405157631e-06,
      "loss": 0.6887,
      "step": 6680
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.674972234505581e-06,
      "loss": 0.689,
      "step": 6685
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6744965481178026e-06,
      "loss": 0.6877,
      "step": 6690
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6740205815500236e-06,
      "loss": 0.6872,
      "step": 6695
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.673544334999957e-06,
      "loss": 0.6999,
      "step": 6700
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6730678086654306e-06,
      "loss": 0.6983,
      "step": 6705
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6725910027443902e-06,
      "loss": 0.6488,
      "step": 6710
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6721139174348964e-06,
      "loss": 0.6741,
      "step": 6715
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.671636552935126e-06,
      "loss": 0.6809,
      "step": 6720
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6711589094433725e-06,
      "loss": 0.7326,
      "step": 6725
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6706809871580446e-06,
      "loss": 0.6757,
      "step": 6730
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.670202786277667e-06,
      "loss": 0.6515,
      "step": 6735
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6697243070008805e-06,
      "loss": 0.6747,
      "step": 6740
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6692455495264413e-06,
      "loss": 0.6816,
      "step": 6745
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6687665140532209e-06,
      "loss": 0.6918,
      "step": 6750
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6682872007802062e-06,
      "loss": 0.7191,
      "step": 6755
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.6678076099064999e-06,
      "loss": 0.6649,
      "step": 6760
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.66732774163132e-06,
      "loss": 0.6889,
      "step": 6765
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.666847596154e-06,
      "loss": 0.6917,
      "step": 6770
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6663671736739874e-06,
      "loss": 0.6717,
      "step": 6775
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.665886474390846e-06,
      "loss": 0.7177,
      "step": 6780
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6654054985042538e-06,
      "loss": 0.6676,
      "step": 6785
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6649242462140044e-06,
      "loss": 0.708,
      "step": 6790
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6644427177200053e-06,
      "loss": 0.6958,
      "step": 6795
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.66396091322228e-06,
      "loss": 0.6862,
      "step": 6800
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.663478832920965e-06,
      "loss": 0.6462,
      "step": 6805
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6629964770163128e-06,
      "loss": 0.6668,
      "step": 6810
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6625138457086897e-06,
      "loss": 0.6926,
      "step": 6815
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6620309391985767e-06,
      "loss": 0.6942,
      "step": 6820
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.661547757686569e-06,
      "loss": 0.6988,
      "step": 6825
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6610643013733756e-06,
      "loss": 0.6548,
      "step": 6830
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6605805704598206e-06,
      "loss": 0.6827,
      "step": 6835
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.660096565146841e-06,
      "loss": 0.6939,
      "step": 6840
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6596122856354885e-06,
      "loss": 0.728,
      "step": 6845
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6591277321269295e-06,
      "loss": 0.6762,
      "step": 6850
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6586429048224422e-06,
      "loss": 0.6697,
      "step": 6855
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6581578039234203e-06,
      "loss": 0.7086,
      "step": 6860
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6576724296313697e-06,
      "loss": 0.7378,
      "step": 6865
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.657186782147912e-06,
      "loss": 0.6861,
      "step": 6870
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6567008616747797e-06,
      "loss": 0.6749,
      "step": 6875
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6562146684138205e-06,
      "loss": 0.67,
      "step": 6880
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.655728202566995e-06,
      "loss": 0.6561,
      "step": 6885
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6552414643363766e-06,
      "loss": 0.7061,
      "step": 6890
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6547544539241516e-06,
      "loss": 0.6779,
      "step": 6895
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6542671715326209e-06,
      "loss": 0.6927,
      "step": 6900
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.653779617364197e-06,
      "loss": 0.6633,
      "step": 6905
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6532917916214055e-06,
      "loss": 0.6755,
      "step": 6910
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6528036945068852e-06,
      "loss": 0.6774,
      "step": 6915
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.652315326223387e-06,
      "loss": 0.7033,
      "step": 6920
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6518266869737754e-06,
      "loss": 0.6823,
      "step": 6925
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6513377769610264e-06,
      "loss": 0.6952,
      "step": 6930
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6508485963882293e-06,
      "loss": 0.676,
      "step": 6935
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.650359145458585e-06,
      "loss": 0.7026,
      "step": 6940
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6498694243754075e-06,
      "loss": 0.6872,
      "step": 6945
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6493794333421228e-06,
      "loss": 0.6846,
      "step": 6950
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6488891725622688e-06,
      "loss": 0.6825,
      "step": 6955
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6483986422394955e-06,
      "loss": 0.723,
      "step": 6960
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6479078425775653e-06,
      "loss": 0.7463,
      "step": 6965
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6474167737803514e-06,
      "loss": 0.7006,
      "step": 6970
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.64692543605184e-06,
      "loss": 0.6745,
      "step": 6975
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6464338295961283e-06,
      "loss": 0.6935,
      "step": 6980
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6459419546174253e-06,
      "loss": 0.7056,
      "step": 6985
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6454498113200521e-06,
      "loss": 0.6763,
      "step": 6990
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6449573999084404e-06,
      "loss": 0.68,
      "step": 6995
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6444647205871332e-06,
      "loss": 0.6595,
      "step": 7000
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6439717735607856e-06,
      "loss": 0.6996,
      "step": 7005
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.643478559034164e-06,
      "loss": 0.7008,
      "step": 7010
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6429850772121446e-06,
      "loss": 0.6907,
      "step": 7015
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6424913282997159e-06,
      "loss": 0.6627,
      "step": 7020
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.641997312501977e-06,
      "loss": 0.6939,
      "step": 7025
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6415030300241371e-06,
      "loss": 0.7002,
      "step": 7030
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6410084810715177e-06,
      "loss": 0.7098,
      "step": 7035
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6405136658495496e-06,
      "loss": 0.7114,
      "step": 7040
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.640018584563775e-06,
      "loss": 0.6951,
      "step": 7045
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.639523237419846e-06,
      "loss": 0.7112,
      "step": 7050
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6390276246235257e-06,
      "loss": 0.6827,
      "step": 7055
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6385317463806878e-06,
      "loss": 0.7133,
      "step": 7060
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6380356028973152e-06,
      "loss": 0.6766,
      "step": 7065
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6375391943795015e-06,
      "loss": 0.6767,
      "step": 7070
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6370425210334514e-06,
      "loss": 0.6731,
      "step": 7075
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6365455830654775e-06,
      "loss": 0.6974,
      "step": 7080
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6360483806820043e-06,
      "loss": 0.6972,
      "step": 7085
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.635550914089565e-06,
      "loss": 0.6974,
      "step": 7090
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.635053183494803e-06,
      "loss": 0.6649,
      "step": 7095
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6345551891044713e-06,
      "loss": 0.6754,
      "step": 7100
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6340569311254323e-06,
      "loss": 0.6877,
      "step": 7105
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6335584097646585e-06,
      "loss": 0.6841,
      "step": 7110
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6330596252292309e-06,
      "loss": 0.6675,
      "step": 7115
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.63256057772634e-06,
      "loss": 0.6743,
      "step": 7120
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6320612674632864e-06,
      "loss": 0.7045,
      "step": 7125
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.631561694647479e-06,
      "loss": 0.6891,
      "step": 7130
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6310618594864355e-06,
      "loss": 0.6722,
      "step": 7135
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6305617621877841e-06,
      "loss": 0.7012,
      "step": 7140
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6300614029592602e-06,
      "loss": 0.665,
      "step": 7145
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6295607820087084e-06,
      "loss": 0.6967,
      "step": 7150
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6290598995440835e-06,
      "loss": 0.7089,
      "step": 7155
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.628558755773446e-06,
      "loss": 0.7176,
      "step": 7160
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6280573509049679e-06,
      "loss": 0.6847,
      "step": 7165
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6275556851469284e-06,
      "loss": 0.6968,
      "step": 7170
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6270537587077145e-06,
      "loss": 0.6586,
      "step": 7175
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6265515717958222e-06,
      "loss": 0.7039,
      "step": 7180
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6260491246198563e-06,
      "loss": 0.725,
      "step": 7185
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6255464173885275e-06,
      "loss": 0.7166,
      "step": 7190
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6250434503106578e-06,
      "loss": 0.67,
      "step": 7195
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6245402235951742e-06,
      "loss": 0.6697,
      "step": 7200
    },
    {
      "epoch": 0.29,
      "eval_loss": 0.6546275019645691,
      "eval_runtime": 138.8622,
      "eval_samples_per_second": 17.038,
      "eval_steps_per_second": 2.845,
      "step": 7200
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.624036737451113e-06,
      "loss": 0.6907,
      "step": 7205
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.623532992087618e-06,
      "loss": 0.7066,
      "step": 7210
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6230289877139403e-06,
      "loss": 0.7354,
      "step": 7215
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6225247245394393e-06,
      "loss": 0.672,
      "step": 7220
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.622020202773582e-06,
      "loss": 0.676,
      "step": 7225
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6215154226259414e-06,
      "loss": 0.6798,
      "step": 7230
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.621010384306199e-06,
      "loss": 0.689,
      "step": 7235
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.620505088024144e-06,
      "loss": 0.7003,
      "step": 7240
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.619999533989671e-06,
      "loss": 0.6669,
      "step": 7245
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6194937224127837e-06,
      "loss": 0.6849,
      "step": 7250
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6189876535035919e-06,
      "loss": 0.6997,
      "step": 7255
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6184813274723113e-06,
      "loss": 0.7292,
      "step": 7260
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6179747445292659e-06,
      "loss": 0.6555,
      "step": 7265
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6174679048848856e-06,
      "loss": 0.672,
      "step": 7270
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6169608087497077e-06,
      "loss": 0.6803,
      "step": 7275
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6164534563343752e-06,
      "loss": 0.6701,
      "step": 7280
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.615945847849638e-06,
      "loss": 0.6493,
      "step": 7285
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.615437983506352e-06,
      "loss": 0.7276,
      "step": 7290
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6149298635154795e-06,
      "loss": 0.6832,
      "step": 7295
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6144214880880895e-06,
      "loss": 0.6981,
      "step": 7300
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6139128574353568e-06,
      "loss": 0.6715,
      "step": 7305
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.613403971768562e-06,
      "loss": 0.6732,
      "step": 7310
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6128948312990916e-06,
      "loss": 0.6906,
      "step": 7315
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6123854362384384e-06,
      "loss": 0.682,
      "step": 7320
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6118757867982002e-06,
      "loss": 0.701,
      "step": 7325
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6113658831900816e-06,
      "loss": 0.6969,
      "step": 7330
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6108557256258916e-06,
      "loss": 0.6774,
      "step": 7335
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6103453143175458e-06,
      "loss": 0.6849,
      "step": 7340
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6098346494770642e-06,
      "loss": 0.6714,
      "step": 7345
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6093237313165722e-06,
      "loss": 0.6781,
      "step": 7350
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6088125600483014e-06,
      "loss": 0.657,
      "step": 7355
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6083011358845878e-06,
      "loss": 0.6867,
      "step": 7360
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6077894590378722e-06,
      "loss": 0.6977,
      "step": 7365
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.607277529720701e-06,
      "loss": 0.6775,
      "step": 7370
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6067653481457251e-06,
      "loss": 0.6962,
      "step": 7375
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6062529145257e-06,
      "loss": 0.6801,
      "step": 7380
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6057402290734867e-06,
      "loss": 0.6917,
      "step": 7385
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6052272920020502e-06,
      "loss": 0.6858,
      "step": 7390
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6047141035244596e-06,
      "loss": 0.689,
      "step": 7395
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6042006638538893e-06,
      "loss": 0.684,
      "step": 7400
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6036869732036175e-06,
      "loss": 0.6685,
      "step": 7405
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.603173031787027e-06,
      "loss": 0.7001,
      "step": 7410
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.602658839817605e-06,
      "loss": 0.6947,
      "step": 7415
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6021443975089415e-06,
      "loss": 0.6384,
      "step": 7420
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.601629705074732e-06,
      "loss": 0.6828,
      "step": 7425
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6011147627287746e-06,
      "loss": 0.6856,
      "step": 7430
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6005995706849726e-06,
      "loss": 0.6655,
      "step": 7435
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6000841291573322e-06,
      "loss": 0.7021,
      "step": 7440
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.599568438359963e-06,
      "loss": 0.6889,
      "step": 7445
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.5990524985070785e-06,
      "loss": 0.6917,
      "step": 7450
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.598536309812996e-06,
      "loss": 0.7008,
      "step": 7455
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.598019872492135e-06,
      "loss": 0.6614,
      "step": 7460
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.59750318675902e-06,
      "loss": 0.6887,
      "step": 7465
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.5969862528282771e-06,
      "loss": 0.6733,
      "step": 7470
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.5964690709146367e-06,
      "loss": 0.6935,
      "step": 7475
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.5959516412329314e-06,
      "loss": 0.6731,
      "step": 7480
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.5954339639980967e-06,
      "loss": 0.6846,
      "step": 7485
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.5949160394251718e-06,
      "loss": 0.7281,
      "step": 7490
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.5943978677292976e-06,
      "loss": 0.7238,
      "step": 7495
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.593879449125718e-06,
      "loss": 0.6682,
      "step": 7500
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.5933607838297804e-06,
      "loss": 0.6864,
      "step": 7505
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5928418720569332e-06,
      "loss": 0.6775,
      "step": 7510
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5923227140227278e-06,
      "loss": 0.7329,
      "step": 7515
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5918033099428182e-06,
      "loss": 0.6964,
      "step": 7520
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.59128366003296e-06,
      "loss": 0.6862,
      "step": 7525
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5907637645090117e-06,
      "loss": 0.6552,
      "step": 7530
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5902436235869333e-06,
      "loss": 0.6738,
      "step": 7535
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5897232374827862e-06,
      "loss": 0.7062,
      "step": 7540
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.589202606412735e-06,
      "loss": 0.6975,
      "step": 7545
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5886817305930452e-06,
      "loss": 0.6795,
      "step": 7550
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5881606102400836e-06,
      "loss": 0.6633,
      "step": 7555
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5876392455703198e-06,
      "loss": 0.6959,
      "step": 7560
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5871176368003231e-06,
      "loss": 0.6719,
      "step": 7565
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5865957841467666e-06,
      "loss": 0.7142,
      "step": 7570
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5860736878264222e-06,
      "loss": 0.6729,
      "step": 7575
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5855513480561649e-06,
      "loss": 0.6777,
      "step": 7580
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5850287650529698e-06,
      "loss": 0.6729,
      "step": 7585
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5845059390339134e-06,
      "loss": 0.7104,
      "step": 7590
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5839828702161727e-06,
      "loss": 0.6844,
      "step": 7595
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5834595588170266e-06,
      "loss": 0.6837,
      "step": 7600
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5829360050538537e-06,
      "loss": 0.7076,
      "step": 7605
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5824122091441337e-06,
      "loss": 0.6906,
      "step": 7610
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5818881713054469e-06,
      "loss": 0.7267,
      "step": 7615
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5813638917554742e-06,
      "loss": 0.7293,
      "step": 7620
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5808393707119967e-06,
      "loss": 0.693,
      "step": 7625
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5803146083928956e-06,
      "loss": 0.7122,
      "step": 7630
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.579789605016153e-06,
      "loss": 0.6895,
      "step": 7635
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5792643607998506e-06,
      "loss": 0.6916,
      "step": 7640
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.57873887596217e-06,
      "loss": 0.6745,
      "step": 7645
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5782131507213934e-06,
      "loss": 0.7242,
      "step": 7650
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5776871852959026e-06,
      "loss": 0.714,
      "step": 7655
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5771609799041788e-06,
      "loss": 0.7086,
      "step": 7660
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.576634534764803e-06,
      "loss": 0.738,
      "step": 7665
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5761078500964562e-06,
      "loss": 0.669,
      "step": 7670
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5755809261179185e-06,
      "loss": 0.6587,
      "step": 7675
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5750537630480696e-06,
      "loss": 0.676,
      "step": 7680
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5745263611058886e-06,
      "loss": 0.6838,
      "step": 7685
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5739987205104535e-06,
      "loss": 0.6182,
      "step": 7690
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5734708414809415e-06,
      "loss": 0.6497,
      "step": 7695
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.572942724236629e-06,
      "loss": 0.6681,
      "step": 7700
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5724143689968915e-06,
      "loss": 0.7043,
      "step": 7705
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5718857759812033e-06,
      "loss": 0.6924,
      "step": 7710
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.571356945409137e-06,
      "loss": 0.6852,
      "step": 7715
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.570827877500364e-06,
      "loss": 0.7008,
      "step": 7720
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5702985724746552e-06,
      "loss": 0.7029,
      "step": 7725
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5697690305518787e-06,
      "loss": 0.6684,
      "step": 7730
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.569239251952002e-06,
      "loss": 0.7089,
      "step": 7735
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5687092368950908e-06,
      "loss": 0.6756,
      "step": 7740
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5681789856013076e-06,
      "loss": 0.6874,
      "step": 7745
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5676484982909154e-06,
      "loss": 0.7055,
      "step": 7750
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.5671177751842733e-06,
      "loss": 0.7159,
      "step": 7755
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5665868165018395e-06,
      "loss": 0.6968,
      "step": 7760
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.566055622464169e-06,
      "loss": 0.6533,
      "step": 7765
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.565524193291916e-06,
      "loss": 0.6589,
      "step": 7770
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.564992529205831e-06,
      "loss": 0.7165,
      "step": 7775
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5644606304267627e-06,
      "loss": 0.6588,
      "step": 7780
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5639284971756574e-06,
      "loss": 0.7004,
      "step": 7785
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5633961296735585e-06,
      "loss": 0.7019,
      "step": 7790
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.562863528141607e-06,
      "loss": 0.6711,
      "step": 7795
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5623306928010408e-06,
      "loss": 0.6558,
      "step": 7800
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.561797623873195e-06,
      "loss": 0.7162,
      "step": 7805
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5612643215795017e-06,
      "loss": 0.6986,
      "step": 7810
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5607307861414905e-06,
      "loss": 0.717,
      "step": 7815
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.560197017780787e-06,
      "loss": 0.7249,
      "step": 7820
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5596630167191138e-06,
      "loss": 0.6965,
      "step": 7825
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5591287831782908e-06,
      "loss": 0.679,
      "step": 7830
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5585943173802333e-06,
      "loss": 0.6812,
      "step": 7835
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5580596195469547e-06,
      "loss": 0.6957,
      "step": 7840
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5575246899005629e-06,
      "loss": 0.6792,
      "step": 7845
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5569895286632634e-06,
      "loss": 0.6675,
      "step": 7850
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5564541360573578e-06,
      "loss": 0.6931,
      "step": 7855
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5559185123052427e-06,
      "loss": 0.7044,
      "step": 7860
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5553826576294127e-06,
      "loss": 0.6978,
      "step": 7865
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5548465722524561e-06,
      "loss": 0.693,
      "step": 7870
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.554310256397059e-06,
      "loss": 0.7044,
      "step": 7875
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5537737102860015e-06,
      "loss": 0.6841,
      "step": 7880
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5532369341421609e-06,
      "loss": 0.674,
      "step": 7885
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5526999281885088e-06,
      "loss": 0.6544,
      "step": 7890
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.552162692648113e-06,
      "loss": 0.6951,
      "step": 7895
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.551625227744137e-06,
      "loss": 0.6969,
      "step": 7900
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5510875336998382e-06,
      "loss": 0.6827,
      "step": 7905
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5505496107385704e-06,
      "loss": 0.6296,
      "step": 7910
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.550011459083782e-06,
      "loss": 0.6884,
      "step": 7915
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.549473078959017e-06,
      "loss": 0.6726,
      "step": 7920
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.548934470587913e-06,
      "loss": 0.6732,
      "step": 7925
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.548395634194204e-06,
      "loss": 0.6986,
      "step": 7930
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5478565700017174e-06,
      "loss": 0.7457,
      "step": 7935
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.547317278234376e-06,
      "loss": 0.713,
      "step": 7940
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5467777591161973e-06,
      "loss": 0.6842,
      "step": 7945
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5462380128712921e-06,
      "loss": 0.7125,
      "step": 7950
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.545698039723867e-06,
      "loss": 0.7007,
      "step": 7955
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5451578398982216e-06,
      "loss": 0.6896,
      "step": 7960
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5446174136187503e-06,
      "loss": 0.7371,
      "step": 7965
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.544076761109942e-06,
      "loss": 0.6922,
      "step": 7970
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5435358825963784e-06,
      "loss": 0.7154,
      "step": 7975
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.542994778302736e-06,
      "loss": 0.7194,
      "step": 7980
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5424534484537847e-06,
      "loss": 0.6813,
      "step": 7985
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5419118932743883e-06,
      "loss": 0.7203,
      "step": 7990
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5413701129895045e-06,
      "loss": 0.6675,
      "step": 7995
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5408281078241835e-06,
      "loss": 0.7152,
      "step": 8000
    },
    {
      "epoch": 0.32,
      "eval_loss": 0.6539024114608765,
      "eval_runtime": 138.566,
      "eval_samples_per_second": 17.075,
      "eval_steps_per_second": 2.851,
      "step": 8000
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5402858780035697e-06,
      "loss": 0.6859,
      "step": 8005
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5397434237529012e-06,
      "loss": 0.7315,
      "step": 8010
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5392007452975077e-06,
      "loss": 0.7081,
      "step": 8015
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5386578428628142e-06,
      "loss": 0.682,
      "step": 8020
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5381147166743369e-06,
      "loss": 0.6741,
      "step": 8025
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5375713669576857e-06,
      "loss": 0.6979,
      "step": 8030
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5370277939385644e-06,
      "loss": 0.6927,
      "step": 8035
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.536483997842767e-06,
      "loss": 0.6856,
      "step": 8040
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5359399788961826e-06,
      "loss": 0.6683,
      "step": 8045
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5353957373247917e-06,
      "loss": 0.6646,
      "step": 8050
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5348512733546674e-06,
      "loss": 0.6684,
      "step": 8055
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5343065872119759e-06,
      "loss": 0.6741,
      "step": 8060
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5337616791229744e-06,
      "loss": 0.6936,
      "step": 8065
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5332165493140133e-06,
      "loss": 0.6646,
      "step": 8070
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5326711980115343e-06,
      "loss": 0.6936,
      "step": 8075
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5321256254420724e-06,
      "loss": 0.7009,
      "step": 8080
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5315798318322532e-06,
      "loss": 0.7183,
      "step": 8085
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5310338174087946e-06,
      "loss": 0.6853,
      "step": 8090
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5304875823985066e-06,
      "loss": 0.6851,
      "step": 8095
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5299411270282898e-06,
      "loss": 0.6607,
      "step": 8100
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5293944515251376e-06,
      "loss": 0.6715,
      "step": 8105
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5288475561161342e-06,
      "loss": 0.6525,
      "step": 8110
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5283004410284549e-06,
      "loss": 0.6958,
      "step": 8115
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5277531064893669e-06,
      "loss": 0.7071,
      "step": 8120
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5272055527262278e-06,
      "loss": 0.683,
      "step": 8125
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.526657779966487e-06,
      "loss": 0.71,
      "step": 8130
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5261097884376848e-06,
      "loss": 0.683,
      "step": 8135
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5255615783674512e-06,
      "loss": 0.6911,
      "step": 8140
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5250131499835088e-06,
      "loss": 0.7089,
      "step": 8145
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5244645035136694e-06,
      "loss": 0.6528,
      "step": 8150
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5239156391858363e-06,
      "loss": 0.6626,
      "step": 8155
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.523366557228003e-06,
      "loss": 0.6674,
      "step": 8160
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5228172578682531e-06,
      "loss": 0.6858,
      "step": 8165
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5222677413347612e-06,
      "loss": 0.6944,
      "step": 8170
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.521718007855791e-06,
      "loss": 0.6623,
      "step": 8175
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5211680576596976e-06,
      "loss": 0.6764,
      "step": 8180
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5206178909749254e-06,
      "loss": 0.7022,
      "step": 8185
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5200675080300086e-06,
      "loss": 0.6437,
      "step": 8190
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.519516909053572e-06,
      "loss": 0.6888,
      "step": 8195
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.518966094274329e-06,
      "loss": 0.6837,
      "step": 8200
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.518415063921084e-06,
      "loss": 0.7228,
      "step": 8205
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5178638182227292e-06,
      "loss": 0.6647,
      "step": 8210
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5173123574082482e-06,
      "loss": 0.6992,
      "step": 8215
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5167606817067129e-06,
      "loss": 0.6828,
      "step": 8220
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5162087913472844e-06,
      "loss": 0.7303,
      "step": 8225
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5156566865592128e-06,
      "loss": 0.6794,
      "step": 8230
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5151043675718383e-06,
      "loss": 0.6442,
      "step": 8235
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5145518346145887e-06,
      "loss": 0.6669,
      "step": 8240
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5139990879169822e-06,
      "loss": 0.6902,
      "step": 8245
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.513446127708624e-06,
      "loss": 0.6503,
      "step": 8250
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5128929542192102e-06,
      "loss": 0.6971,
      "step": 8255
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.512339567678523e-06,
      "loss": 0.7052,
      "step": 8260
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.511785968316435e-06,
      "loss": 0.6673,
      "step": 8265
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5112321563629066e-06,
      "loss": 0.6816,
      "step": 8270
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5106781320479862e-06,
      "loss": 0.6708,
      "step": 8275
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5101238956018109e-06,
      "loss": 0.7032,
      "step": 8280
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5095694472546058e-06,
      "loss": 0.7008,
      "step": 8285
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5090147872366835e-06,
      "loss": 0.719,
      "step": 8290
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5084599157784453e-06,
      "loss": 0.7094,
      "step": 8295
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5079048331103797e-06,
      "loss": 0.7306,
      "step": 8300
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5073495394630636e-06,
      "loss": 0.7004,
      "step": 8305
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5067940350671606e-06,
      "loss": 0.7107,
      "step": 8310
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.506238320153423e-06,
      "loss": 0.7033,
      "step": 8315
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5056823949526898e-06,
      "loss": 0.6855,
      "step": 8320
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5051262596958872e-06,
      "loss": 0.7119,
      "step": 8325
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5045699146140289e-06,
      "loss": 0.6853,
      "step": 8330
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5040133599382162e-06,
      "loss": 0.71,
      "step": 8335
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.503456595899637e-06,
      "loss": 0.6527,
      "step": 8340
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5028996227295664e-06,
      "loss": 0.6692,
      "step": 8345
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5023424406593654e-06,
      "loss": 0.6568,
      "step": 8350
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5017850499204835e-06,
      "loss": 0.658,
      "step": 8355
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.501227450744455e-06,
      "loss": 0.6895,
      "step": 8360
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5006696433629032e-06,
      "loss": 0.7062,
      "step": 8365
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5001116280075353e-06,
      "loss": 0.6883,
      "step": 8370
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.499553404910146e-06,
      "loss": 0.666,
      "step": 8375
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.4989949743026169e-06,
      "loss": 0.709,
      "step": 8380
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.4984363364169145e-06,
      "loss": 0.7225,
      "step": 8385
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.4978774914850933e-06,
      "loss": 0.6657,
      "step": 8390
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.4973184397392915e-06,
      "loss": 0.6774,
      "step": 8395
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.4967591814117347e-06,
      "loss": 0.6617,
      "step": 8400
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.496199716734734e-06,
      "loss": 0.669,
      "step": 8405
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.4956400459406862e-06,
      "loss": 0.6798,
      "step": 8410
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.4950801692620735e-06,
      "loss": 0.6421,
      "step": 8415
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.494520086931464e-06,
      "loss": 0.6742,
      "step": 8420
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.4939597991815107e-06,
      "loss": 0.6478,
      "step": 8425
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.493399306244953e-06,
      "loss": 0.6817,
      "step": 8430
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.492838608354614e-06,
      "loss": 0.7469,
      "step": 8435
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.4922777057434031e-06,
      "loss": 0.6713,
      "step": 8440
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.4917165986443142e-06,
      "loss": 0.7093,
      "step": 8445
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.4911552872904266e-06,
      "loss": 0.6825,
      "step": 8450
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.4905937719149035e-06,
      "loss": 0.664,
      "step": 8455
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.4900320527509942e-06,
      "loss": 0.6827,
      "step": 8460
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.489470130032032e-06,
      "loss": 0.7093,
      "step": 8465
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.488908003991434e-06,
      "loss": 0.7162,
      "step": 8470
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.4883456748627032e-06,
      "loss": 0.6738,
      "step": 8475
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.4877831428794258e-06,
      "loss": 0.6977,
      "step": 8480
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.4872204082752728e-06,
      "loss": 0.7097,
      "step": 8485
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.4866574712839994e-06,
      "loss": 0.6716,
      "step": 8490
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4860943321394443e-06,
      "loss": 0.6891,
      "step": 8495
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4855309910755313e-06,
      "loss": 0.6471,
      "step": 8500
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4849674483262668e-06,
      "loss": 0.6544,
      "step": 8505
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4844037041257416e-06,
      "loss": 0.6554,
      "step": 8510
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4838397587081307e-06,
      "loss": 0.694,
      "step": 8515
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4832756123076912e-06,
      "loss": 0.6488,
      "step": 8520
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4827112651587656e-06,
      "loss": 0.6885,
      "step": 8525
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.482146717495778e-06,
      "loss": 0.6827,
      "step": 8530
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.481581969553237e-06,
      "loss": 0.7017,
      "step": 8535
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.481017021565734e-06,
      "loss": 0.681,
      "step": 8540
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4804518737679432e-06,
      "loss": 0.6228,
      "step": 8545
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4798865263946223e-06,
      "loss": 0.7208,
      "step": 8550
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4793209796806117e-06,
      "loss": 0.6627,
      "step": 8555
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4787552338608341e-06,
      "loss": 0.6698,
      "step": 8560
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4781892891702965e-06,
      "loss": 0.6631,
      "step": 8565
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4776231458440862e-06,
      "loss": 0.6991,
      "step": 8570
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.477056804117375e-06,
      "loss": 0.65,
      "step": 8575
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.476490264225416e-06,
      "loss": 0.6846,
      "step": 8580
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.475923526403545e-06,
      "loss": 0.6916,
      "step": 8585
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.47535659088718e-06,
      "loss": 0.654,
      "step": 8590
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4747894579118208e-06,
      "loss": 0.7077,
      "step": 8595
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.47422212771305e-06,
      "loss": 0.6887,
      "step": 8600
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4736546005265314e-06,
      "loss": 0.6919,
      "step": 8605
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4730868765880109e-06,
      "loss": 0.6471,
      "step": 8610
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4725189561333158e-06,
      "loss": 0.706,
      "step": 8615
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4719508393983555e-06,
      "loss": 0.6916,
      "step": 8620
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.471382526619121e-06,
      "loss": 0.6624,
      "step": 8625
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4708140180316843e-06,
      "loss": 0.6888,
      "step": 8630
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.470245313872199e-06,
      "loss": 0.6796,
      "step": 8635
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4696764143768997e-06,
      "loss": 0.6776,
      "step": 8640
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.469107319782102e-06,
      "loss": 0.7217,
      "step": 8645
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4685380303242037e-06,
      "loss": 0.6858,
      "step": 8650
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4679685462396817e-06,
      "loss": 0.6937,
      "step": 8655
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.467398867765096e-06,
      "loss": 0.6747,
      "step": 8660
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4668289951370848e-06,
      "loss": 0.6699,
      "step": 8665
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4662589285923686e-06,
      "loss": 0.7111,
      "step": 8670
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4656886683677486e-06,
      "loss": 0.667,
      "step": 8675
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4651182147001055e-06,
      "loss": 0.6586,
      "step": 8680
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.464547567826401e-06,
      "loss": 0.6996,
      "step": 8685
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4639767279836766e-06,
      "loss": 0.7181,
      "step": 8690
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.463405695409054e-06,
      "loss": 0.6878,
      "step": 8695
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.462834470339736e-06,
      "loss": 0.6247,
      "step": 8700
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4622630530130037e-06,
      "loss": 0.7024,
      "step": 8705
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4616914436662195e-06,
      "loss": 0.6748,
      "step": 8710
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4611196425368247e-06,
      "loss": 0.6877,
      "step": 8715
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.46054764986234e-06,
      "loss": 0.663,
      "step": 8720
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4599754658803671e-06,
      "loss": 0.6649,
      "step": 8725
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4594030908285858e-06,
      "loss": 0.6701,
      "step": 8730
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4588305249447557e-06,
      "loss": 0.6684,
      "step": 8735
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4582577684667156e-06,
      "loss": 0.7104,
      "step": 8740
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.457684821632384e-06,
      "loss": 0.7183,
      "step": 8745
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.457111684679757e-06,
      "loss": 0.7016,
      "step": 8750
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4565383578469119e-06,
      "loss": 0.6959,
      "step": 8755
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4559648413720033e-06,
      "loss": 0.7299,
      "step": 8760
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4553911354932646e-06,
      "loss": 0.6539,
      "step": 8765
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4548172404490089e-06,
      "loss": 0.685,
      "step": 8770
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4542431564776265e-06,
      "loss": 0.7004,
      "step": 8775
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.453668883817587e-06,
      "loss": 0.6742,
      "step": 8780
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.453094422707439e-06,
      "loss": 0.6744,
      "step": 8785
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4525197733858077e-06,
      "loss": 0.6871,
      "step": 8790
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.451944936091398e-06,
      "loss": 0.6567,
      "step": 8795
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4513699110629921e-06,
      "loss": 0.6838,
      "step": 8800
    },
    {
      "epoch": 0.36,
      "eval_loss": 0.6491459012031555,
      "eval_runtime": 138.8275,
      "eval_samples_per_second": 17.043,
      "eval_steps_per_second": 2.845,
      "step": 8800
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.45079469853945e-06,
      "loss": 0.6431,
      "step": 8805
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4502192987597113e-06,
      "loss": 0.6606,
      "step": 8810
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4496437119627905e-06,
      "loss": 0.6917,
      "step": 8815
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4490679383877825e-06,
      "loss": 0.6826,
      "step": 8820
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4484919782738581e-06,
      "loss": 0.6904,
      "step": 8825
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4479158318602658e-06,
      "loss": 0.6781,
      "step": 8830
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4473394993863325e-06,
      "loss": 0.7099,
      "step": 8835
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4467629810914615e-06,
      "loss": 0.6993,
      "step": 8840
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4461862772151333e-06,
      "loss": 0.6606,
      "step": 8845
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4456093879969057e-06,
      "loss": 0.7129,
      "step": 8850
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4450323136764136e-06,
      "loss": 0.6789,
      "step": 8855
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4444550544933684e-06,
      "loss": 0.6828,
      "step": 8860
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.443877610687559e-06,
      "loss": 0.701,
      "step": 8865
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4432999824988503e-06,
      "loss": 0.6962,
      "step": 8870
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.442722170167184e-06,
      "loss": 0.6399,
      "step": 8875
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4421441739325782e-06,
      "loss": 0.6943,
      "step": 8880
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4415659940351275e-06,
      "loss": 0.6932,
      "step": 8885
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.440987630715003e-06,
      "loss": 0.7096,
      "step": 8890
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4404090842124519e-06,
      "loss": 0.6652,
      "step": 8895
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.439830354767797e-06,
      "loss": 0.6988,
      "step": 8900
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4392514426214378e-06,
      "loss": 0.6862,
      "step": 8905
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4386723480138491e-06,
      "loss": 0.7313,
      "step": 8910
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.438093071185582e-06,
      "loss": 0.6876,
      "step": 8915
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.437513612377263e-06,
      "loss": 0.664,
      "step": 8920
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.436933971829594e-06,
      "loss": 0.6918,
      "step": 8925
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4363541497833534e-06,
      "loss": 0.7093,
      "step": 8930
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4357741464793932e-06,
      "loss": 0.7008,
      "step": 8935
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4351939621586424e-06,
      "loss": 0.6459,
      "step": 8940
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4346135970621045e-06,
      "loss": 0.7083,
      "step": 8945
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4340330514308576e-06,
      "loss": 0.6674,
      "step": 8950
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4334523255060563e-06,
      "loss": 0.6906,
      "step": 8955
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.432871419528928e-06,
      "loss": 0.6976,
      "step": 8960
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.432290333740776e-06,
      "loss": 0.6487,
      "step": 8965
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4317090683829797e-06,
      "loss": 0.6716,
      "step": 8970
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.43112762369699e-06,
      "loss": 0.6578,
      "step": 8975
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4305459999243353e-06,
      "loss": 0.6437,
      "step": 8980
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.429964197306616e-06,
      "loss": 0.6527,
      "step": 8985
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4293822160855083e-06,
      "loss": 0.7384,
      "step": 8990
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4288000565027623e-06,
      "loss": 0.6817,
      "step": 8995
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4282177188002016e-06,
      "loss": 0.6805,
      "step": 9000
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.427635203219725e-06,
      "loss": 0.6908,
      "step": 9005
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4270525100033036e-06,
      "loss": 0.6745,
      "step": 9010
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4264696393929832e-06,
      "loss": 0.6801,
      "step": 9015
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4258865916308834e-06,
      "loss": 0.6856,
      "step": 9020
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4253033669591971e-06,
      "loss": 0.6676,
      "step": 9025
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.424719965620191e-06,
      "loss": 0.6524,
      "step": 9030
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4241363878562046e-06,
      "loss": 0.6867,
      "step": 9035
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4235526339096514e-06,
      "loss": 0.6674,
      "step": 9040
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.422968704023017e-06,
      "loss": 0.659,
      "step": 9045
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4223845984388613e-06,
      "loss": 0.6706,
      "step": 9050
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.421800317399817e-06,
      "loss": 0.6867,
      "step": 9055
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.421215861148589e-06,
      "loss": 0.7136,
      "step": 9060
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.420631229927955e-06,
      "loss": 0.7003,
      "step": 9065
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4200464239807664e-06,
      "loss": 0.6764,
      "step": 9070
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4194614435499458e-06,
      "loss": 0.6954,
      "step": 9075
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4188762888784897e-06,
      "loss": 0.6641,
      "step": 9080
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4182909602094662e-06,
      "loss": 0.6822,
      "step": 9085
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.417705457786015e-06,
      "loss": 0.7138,
      "step": 9090
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4171197818513495e-06,
      "loss": 0.7152,
      "step": 9095
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4165339326487535e-06,
      "loss": 0.6324,
      "step": 9100
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4159479104215846e-06,
      "loss": 0.6677,
      "step": 9105
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4153617154132712e-06,
      "loss": 0.6521,
      "step": 9110
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.414775347867313e-06,
      "loss": 0.7191,
      "step": 9115
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4141888080272825e-06,
      "loss": 0.6545,
      "step": 9120
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4136020961368228e-06,
      "loss": 0.7033,
      "step": 9125
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4130152124396497e-06,
      "loss": 0.6917,
      "step": 9130
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.412428157179549e-06,
      "loss": 0.7083,
      "step": 9135
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4118409306003784e-06,
      "loss": 0.6786,
      "step": 9140
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4112535329460671e-06,
      "loss": 0.677,
      "step": 9145
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4106659644606139e-06,
      "loss": 0.6903,
      "step": 9150
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.410078225388091e-06,
      "loss": 0.659,
      "step": 9155
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.409490315972639e-06,
      "loss": 0.7028,
      "step": 9160
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4089022364584712e-06,
      "loss": 0.6607,
      "step": 9165
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4083139870898706e-06,
      "loss": 0.6595,
      "step": 9170
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4077255681111903e-06,
      "loss": 0.6557,
      "step": 9175
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4071369797668545e-06,
      "loss": 0.659,
      "step": 9180
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4065482223013585e-06,
      "loss": 0.6879,
      "step": 9185
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.405959295959266e-06,
      "loss": 0.6708,
      "step": 9190
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4053702009852128e-06,
      "loss": 0.6839,
      "step": 9195
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4047809376239034e-06,
      "loss": 0.6835,
      "step": 9200
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4041915061201122e-06,
      "loss": 0.7229,
      "step": 9205
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4036019067186843e-06,
      "loss": 0.6944,
      "step": 9210
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4030121396645349e-06,
      "loss": 0.6232,
      "step": 9215
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.402422205202647e-06,
      "loss": 0.7058,
      "step": 9220
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4018321035780747e-06,
      "loss": 0.6815,
      "step": 9225
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4012418350359414e-06,
      "loss": 0.6449,
      "step": 9230
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4006513998214386e-06,
      "loss": 0.6611,
      "step": 9235
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4000607981798292e-06,
      "loss": 0.6489,
      "step": 9240
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3994700303564431e-06,
      "loss": 0.6866,
      "step": 9245
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3988790965966801e-06,
      "loss": 0.6553,
      "step": 9250
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3982879971460096e-06,
      "loss": 0.6936,
      "step": 9255
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3976967322499683e-06,
      "loss": 0.6741,
      "step": 9260
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3971053021541633e-06,
      "loss": 0.6491,
      "step": 9265
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.396513707104269e-06,
      "loss": 0.6819,
      "step": 9270
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.395921947346029e-06,
      "loss": 0.6472,
      "step": 9275
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3953300231252554e-06,
      "loss": 0.6588,
      "step": 9280
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.394737934687828e-06,
      "loss": 0.6795,
      "step": 9285
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3941456822796954e-06,
      "loss": 0.6876,
      "step": 9290
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.393553266146874e-06,
      "loss": 0.6439,
      "step": 9295
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3929606865354484e-06,
      "loss": 0.6678,
      "step": 9300
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3923679436915711e-06,
      "loss": 0.6646,
      "step": 9305
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3917750378614619e-06,
      "loss": 0.6721,
      "step": 9310
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.391181969291409e-06,
      "loss": 0.6741,
      "step": 9315
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3905887382277683e-06,
      "loss": 0.6298,
      "step": 9320
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3899953449169622e-06,
      "loss": 0.6536,
      "step": 9325
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3894017896054814e-06,
      "loss": 0.7037,
      "step": 9330
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3888080725398836e-06,
      "loss": 0.6777,
      "step": 9335
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3882141939667937e-06,
      "loss": 0.6805,
      "step": 9340
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3876201541329038e-06,
      "loss": 0.666,
      "step": 9345
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3870259532849726e-06,
      "loss": 0.6619,
      "step": 9350
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3864315916698264e-06,
      "loss": 0.7109,
      "step": 9355
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3858370695343573e-06,
      "loss": 0.6914,
      "step": 9360
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3852423871255252e-06,
      "loss": 0.6847,
      "step": 9365
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3846475446903555e-06,
      "loss": 0.6819,
      "step": 9370
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.384052542475941e-06,
      "loss": 0.7004,
      "step": 9375
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3834573807294404e-06,
      "loss": 0.7206,
      "step": 9380
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.382862059698078e-06,
      "loss": 0.7003,
      "step": 9385
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3822665796291459e-06,
      "loss": 0.686,
      "step": 9390
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3816709407700007e-06,
      "loss": 0.682,
      "step": 9395
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3810751433680662e-06,
      "loss": 0.6414,
      "step": 9400
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.380479187670831e-06,
      "loss": 0.6494,
      "step": 9405
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.37988307392585e-06,
      "loss": 0.6997,
      "step": 9410
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3792868023807441e-06,
      "loss": 0.6835,
      "step": 9415
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3786903732831982e-06,
      "loss": 0.6819,
      "step": 9420
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.378093786880965e-06,
      "loss": 0.6568,
      "step": 9425
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3774970434218609e-06,
      "loss": 0.6811,
      "step": 9430
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3769001431537675e-06,
      "loss": 0.6692,
      "step": 9435
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3763030863246326e-06,
      "loss": 0.6732,
      "step": 9440
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3757058731824679e-06,
      "loss": 0.6602,
      "step": 9445
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3751085039753508e-06,
      "loss": 0.6597,
      "step": 9450
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3745109789514233e-06,
      "loss": 0.6906,
      "step": 9455
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3739132983588921e-06,
      "loss": 0.6466,
      "step": 9460
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3733154624460283e-06,
      "loss": 0.6613,
      "step": 9465
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.372717471461168e-06,
      "loss": 0.6821,
      "step": 9470
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3721193256527116e-06,
      "loss": 0.6539,
      "step": 9475
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3715210252691232e-06,
      "loss": 0.6994,
      "step": 9480
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3709225705589318e-06,
      "loss": 0.6556,
      "step": 9485
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3703239617707306e-06,
      "loss": 0.7155,
      "step": 9490
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3697251991531756e-06,
      "loss": 0.6621,
      "step": 9495
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3691262829549883e-06,
      "loss": 0.6771,
      "step": 9500
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3685272134249535e-06,
      "loss": 0.7038,
      "step": 9505
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3679279908119187e-06,
      "loss": 0.6903,
      "step": 9510
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3673286153647967e-06,
      "loss": 0.7236,
      "step": 9515
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3667290873325618e-06,
      "loss": 0.6958,
      "step": 9520
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3661294069642537e-06,
      "loss": 0.653,
      "step": 9525
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3655295745089736e-06,
      "loss": 0.6925,
      "step": 9530
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3649295902158873e-06,
      "loss": 0.6607,
      "step": 9535
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3643294543342228e-06,
      "loss": 0.7378,
      "step": 9540
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.363729167113271e-06,
      "loss": 0.6789,
      "step": 9545
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3631287288023866e-06,
      "loss": 0.6862,
      "step": 9550
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.362528139650986e-06,
      "loss": 0.7204,
      "step": 9555
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.361927399908549e-06,
      "loss": 0.6474,
      "step": 9560
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3613265098246176e-06,
      "loss": 0.6564,
      "step": 9565
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3607254696487962e-06,
      "loss": 0.6762,
      "step": 9570
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.360124279630752e-06,
      "loss": 0.6903,
      "step": 9575
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3595229400202134e-06,
      "loss": 0.6778,
      "step": 9580
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3589214510669722e-06,
      "loss": 0.7117,
      "step": 9585
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3583198130208819e-06,
      "loss": 0.6531,
      "step": 9590
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3577180261318569e-06,
      "loss": 0.7066,
      "step": 9595
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.357116090649875e-06,
      "loss": 0.7132,
      "step": 9600
    },
    {
      "epoch": 0.39,
      "eval_loss": 0.6456849575042725,
      "eval_runtime": 138.6043,
      "eval_samples_per_second": 17.07,
      "eval_steps_per_second": 2.85,
      "step": 9600
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3565140068249746e-06,
      "loss": 0.6633,
      "step": 9605
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3559117749072559e-06,
      "loss": 0.6652,
      "step": 9610
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3553093951468817e-06,
      "loss": 0.6863,
      "step": 9615
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.354706867794074e-06,
      "loss": 0.6849,
      "step": 9620
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3541041930991187e-06,
      "loss": 0.6647,
      "step": 9625
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3535013713123606e-06,
      "loss": 0.6668,
      "step": 9630
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.352898402684207e-06,
      "loss": 0.6574,
      "step": 9635
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3522952874651262e-06,
      "loss": 0.6735,
      "step": 9640
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3516920259056467e-06,
      "loss": 0.688,
      "step": 9645
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.351088618256358e-06,
      "loss": 0.6553,
      "step": 9650
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3504850647679104e-06,
      "loss": 0.6818,
      "step": 9655
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.349881365691015e-06,
      "loss": 0.6596,
      "step": 9660
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.349277521276443e-06,
      "loss": 0.6655,
      "step": 9665
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3486735317750257e-06,
      "loss": 0.6465,
      "step": 9670
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3480693974376561e-06,
      "loss": 0.6647,
      "step": 9675
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3474651185152854e-06,
      "loss": 0.6955,
      "step": 9680
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3468606952589265e-06,
      "loss": 0.6851,
      "step": 9685
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.346256127919651e-06,
      "loss": 0.6865,
      "step": 9690
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.345651416748591e-06,
      "loss": 0.6518,
      "step": 9695
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3450465619969386e-06,
      "loss": 0.67,
      "step": 9700
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.344441563915945e-06,
      "loss": 0.7331,
      "step": 9705
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3438364227569215e-06,
      "loss": 0.7002,
      "step": 9710
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3432311387712377e-06,
      "loss": 0.6777,
      "step": 9715
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3426257122103237e-06,
      "loss": 0.7075,
      "step": 9720
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3420201433256689e-06,
      "loss": 0.707,
      "step": 9725
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3414144323688204e-06,
      "loss": 0.67,
      "step": 9730
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3408085795913862e-06,
      "loss": 0.6853,
      "step": 9735
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3402025852450317e-06,
      "loss": 0.6859,
      "step": 9740
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3395964495814815e-06,
      "loss": 0.7109,
      "step": 9745
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.33899017285252e-06,
      "loss": 0.7126,
      "step": 9750
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3383837553099882e-06,
      "loss": 0.6782,
      "step": 9755
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3377771972057876e-06,
      "loss": 0.6577,
      "step": 9760
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3371704987918763e-06,
      "loss": 0.6859,
      "step": 9765
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3365636603202724e-06,
      "loss": 0.705,
      "step": 9770
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3359566820430509e-06,
      "loss": 0.6359,
      "step": 9775
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.335349564212345e-06,
      "loss": 0.6926,
      "step": 9780
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3347423070803465e-06,
      "loss": 0.6667,
      "step": 9785
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3341349108993047e-06,
      "loss": 0.7033,
      "step": 9790
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3335273759215269e-06,
      "loss": 0.6923,
      "step": 9795
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3329197023993774e-06,
      "loss": 0.706,
      "step": 9800
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3323118905852789e-06,
      "loss": 0.6059,
      "step": 9805
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.331703940731711e-06,
      "loss": 0.7087,
      "step": 9810
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3310958530912106e-06,
      "loss": 0.7124,
      "step": 9815
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3304876279163725e-06,
      "loss": 0.6984,
      "step": 9820
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3298792654598474e-06,
      "loss": 0.6993,
      "step": 9825
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3292707659743442e-06,
      "loss": 0.7156,
      "step": 9830
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3286621297126285e-06,
      "loss": 0.7062,
      "step": 9835
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3280533569275223e-06,
      "loss": 0.6437,
      "step": 9840
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3274444478719048e-06,
      "loss": 0.6651,
      "step": 9845
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3268354027987108e-06,
      "loss": 0.6647,
      "step": 9850
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3262262219609329e-06,
      "loss": 0.6911,
      "step": 9855
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3256169056116195e-06,
      "loss": 0.6931,
      "step": 9860
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3250074540038751e-06,
      "loss": 0.6979,
      "step": 9865
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.324397867390861e-06,
      "loss": 0.6642,
      "step": 9870
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3237881460257933e-06,
      "loss": 0.6457,
      "step": 9875
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3231782901619458e-06,
      "loss": 0.6684,
      "step": 9880
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3225683000526474e-06,
      "loss": 0.6744,
      "step": 9885
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.321958175951282e-06,
      "loss": 0.6504,
      "step": 9890
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3213479181112903e-06,
      "loss": 0.7047,
      "step": 9895
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.320737526786168e-06,
      "loss": 0.6802,
      "step": 9900
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.320127002229466e-06,
      "loss": 0.6668,
      "step": 9905
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.319516344694792e-06,
      "loss": 0.6759,
      "step": 9910
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3189055544358065e-06,
      "loss": 0.6614,
      "step": 9915
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3182946317062271e-06,
      "loss": 0.6613,
      "step": 9920
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3176835767598259e-06,
      "loss": 0.7015,
      "step": 9925
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3170723898504293e-06,
      "loss": 0.6607,
      "step": 9930
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3164610712319197e-06,
      "loss": 0.7008,
      "step": 9935
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3158496211582328e-06,
      "loss": 0.6727,
      "step": 9940
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3152380398833604e-06,
      "loss": 0.6481,
      "step": 9945
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3146263276613474e-06,
      "loss": 0.6803,
      "step": 9950
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3140144847462943e-06,
      "loss": 0.6999,
      "step": 9955
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.313402511392355e-06,
      "loss": 0.7099,
      "step": 9960
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3127904078537378e-06,
      "loss": 0.6695,
      "step": 9965
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3121781743847054e-06,
      "loss": 0.6898,
      "step": 9970
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.311565811239574e-06,
      "loss": 0.6724,
      "step": 9975
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3109533186727145e-06,
      "loss": 0.6905,
      "step": 9980
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3103406969385502e-06,
      "loss": 0.6442,
      "step": 9985
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3097279462915594e-06,
      "loss": 0.6735,
      "step": 9990
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3091150669862734e-06,
      "loss": 0.6962,
      "step": 9995
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3085020592772767e-06,
      "loss": 0.7271,
      "step": 10000
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3078889234192074e-06,
      "loss": 0.7095,
      "step": 10005
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3072756596667569e-06,
      "loss": 0.6467,
      "step": 10010
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3066622682746693e-06,
      "loss": 0.6728,
      "step": 10015
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3060487494977426e-06,
      "loss": 0.6724,
      "step": 10020
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3054351035908269e-06,
      "loss": 0.6336,
      "step": 10025
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3048213308088253e-06,
      "loss": 0.677,
      "step": 10030
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3042074314066937e-06,
      "loss": 0.7052,
      "step": 10035
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3035934056394404e-06,
      "loss": 0.6877,
      "step": 10040
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3029792537621269e-06,
      "loss": 0.6824,
      "step": 10045
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.302364976029866e-06,
      "loss": 0.6759,
      "step": 10050
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3017505726978239e-06,
      "loss": 0.7549,
      "step": 10055
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3011360440212178e-06,
      "loss": 0.686,
      "step": 10060
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3005213902553177e-06,
      "loss": 0.6835,
      "step": 10065
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2999066116554457e-06,
      "loss": 0.7057,
      "step": 10070
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2992917084769755e-06,
      "loss": 0.6839,
      "step": 10075
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2986766809753322e-06,
      "loss": 0.676,
      "step": 10080
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.298061529405993e-06,
      "loss": 0.7068,
      "step": 10085
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2974462540244868e-06,
      "loss": 0.7212,
      "step": 10090
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2968308550863932e-06,
      "loss": 0.6873,
      "step": 10095
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2962153328473435e-06,
      "loss": 0.6773,
      "step": 10100
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.295599687563021e-06,
      "loss": 0.6805,
      "step": 10105
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2949839194891587e-06,
      "loss": 0.6523,
      "step": 10110
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2943680288815418e-06,
      "loss": 0.6765,
      "step": 10115
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2937520159960054e-06,
      "loss": 0.6897,
      "step": 10120
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2931358810884362e-06,
      "loss": 0.6637,
      "step": 10125
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2925196244147712e-06,
      "loss": 0.6691,
      "step": 10130
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.291903246230998e-06,
      "loss": 0.6603,
      "step": 10135
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2912867467931551e-06,
      "loss": 0.7122,
      "step": 10140
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2906701263573306e-06,
      "loss": 0.6517,
      "step": 10145
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2900533851796632e-06,
      "loss": 0.7098,
      "step": 10150
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2894365235163425e-06,
      "loss": 0.6925,
      "step": 10155
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2888195416236065e-06,
      "loss": 0.6564,
      "step": 10160
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.288202439757745e-06,
      "loss": 0.6779,
      "step": 10165
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2875852181750964e-06,
      "loss": 0.7095,
      "step": 10170
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2869678771320484e-06,
      "loss": 0.6668,
      "step": 10175
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2863504168850406e-06,
      "loss": 0.6538,
      "step": 10180
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2857328376905593e-06,
      "loss": 0.663,
      "step": 10185
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2851151398051427e-06,
      "loss": 0.648,
      "step": 10190
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2844973234853762e-06,
      "loss": 0.6934,
      "step": 10195
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2838793889878955e-06,
      "loss": 0.6829,
      "step": 10200
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2832613365693857e-06,
      "loss": 0.6484,
      "step": 10205
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2826431664865795e-06,
      "loss": 0.6949,
      "step": 10210
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2820248789962605e-06,
      "loss": 0.6975,
      "step": 10215
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.281406474355259e-06,
      "loss": 0.6907,
      "step": 10220
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2807879528204554e-06,
      "loss": 0.6599,
      "step": 10225
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.280169314648778e-06,
      "loss": 0.6655,
      "step": 10230
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2795505600972036e-06,
      "loss": 0.6827,
      "step": 10235
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2789316894227582e-06,
      "loss": 0.6679,
      "step": 10240
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2783127028825143e-06,
      "loss": 0.6862,
      "step": 10245
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2776936007335942e-06,
      "loss": 0.6659,
      "step": 10250
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2770743832331667e-06,
      "loss": 0.7268,
      "step": 10255
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2764550506384504e-06,
      "loss": 0.6597,
      "step": 10260
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2758356032067103e-06,
      "loss": 0.6416,
      "step": 10265
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2752160411952592e-06,
      "loss": 0.7008,
      "step": 10270
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2745963648614583e-06,
      "loss": 0.6704,
      "step": 10275
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2739765744627154e-06,
      "loss": 0.6444,
      "step": 10280
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2733566702564855e-06,
      "loss": 0.6594,
      "step": 10285
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2727366525002728e-06,
      "loss": 0.6597,
      "step": 10290
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2721165214516259e-06,
      "loss": 0.6624,
      "step": 10295
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2714962773681427e-06,
      "loss": 0.6726,
      "step": 10300
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.270875920507467e-06,
      "loss": 0.6924,
      "step": 10305
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2702554511272894e-06,
      "loss": 0.631,
      "step": 10310
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2696348694853476e-06,
      "loss": 0.6702,
      "step": 10315
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2690141758394259e-06,
      "loss": 0.6808,
      "step": 10320
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2683933704473552e-06,
      "loss": 0.7162,
      "step": 10325
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2677724535670126e-06,
      "loss": 0.6696,
      "step": 10330
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.267151425456321e-06,
      "loss": 0.6856,
      "step": 10335
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2665302863732508e-06,
      "loss": 0.6957,
      "step": 10340
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2659090365758174e-06,
      "loss": 0.6934,
      "step": 10345
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2652876763220828e-06,
      "loss": 0.6666,
      "step": 10350
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2646662058701547e-06,
      "loss": 0.7049,
      "step": 10355
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2640446254781855e-06,
      "loss": 0.685,
      "step": 10360
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.263422935404376e-06,
      "loss": 0.6696,
      "step": 10365
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2628011359069694e-06,
      "loss": 0.6905,
      "step": 10370
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2621792272442569e-06,
      "loss": 0.7034,
      "step": 10375
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2615572096745732e-06,
      "loss": 0.6746,
      "step": 10380
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2609350834562992e-06,
      "loss": 0.6676,
      "step": 10385
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2603128488478605e-06,
      "loss": 0.6716,
      "step": 10390
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2596905061077283e-06,
      "loss": 0.6554,
      "step": 10395
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2590680554944184e-06,
      "loss": 0.6564,
      "step": 10400
    },
    {
      "epoch": 0.42,
      "eval_loss": 0.6432496309280396,
      "eval_runtime": 144.6412,
      "eval_samples_per_second": 16.358,
      "eval_steps_per_second": 2.731,
      "step": 10400
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2584454972664912e-06,
      "loss": 0.6727,
      "step": 10405
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.257822831682552e-06,
      "loss": 0.6545,
      "step": 10410
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2572000590012505e-06,
      "loss": 0.6794,
      "step": 10415
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2565771794812812e-06,
      "loss": 0.682,
      "step": 10420
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.255954193381383e-06,
      "loss": 0.6566,
      "step": 10425
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2553311009603389e-06,
      "loss": 0.6396,
      "step": 10430
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2547079024769756e-06,
      "loss": 0.6685,
      "step": 10435
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2540845981901648e-06,
      "loss": 0.6409,
      "step": 10440
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2534611883588213e-06,
      "loss": 0.6928,
      "step": 10445
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2528376732419048e-06,
      "loss": 0.7162,
      "step": 10450
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2522140530984173e-06,
      "loss": 0.6575,
      "step": 10455
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2515903281874057e-06,
      "loss": 0.7134,
      "step": 10460
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2509664987679599e-06,
      "loss": 0.6521,
      "step": 10465
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2503425650992124e-06,
      "loss": 0.6655,
      "step": 10470
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2497185274403407e-06,
      "loss": 0.6606,
      "step": 10475
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2490943860505645e-06,
      "loss": 0.6419,
      "step": 10480
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2484701411891465e-06,
      "loss": 0.6822,
      "step": 10485
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2478457931153927e-06,
      "loss": 0.6705,
      "step": 10490
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2472213420886518e-06,
      "loss": 0.638,
      "step": 10495
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2465967883683152e-06,
      "loss": 0.6684,
      "step": 10500
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2459721322138173e-06,
      "loss": 0.668,
      "step": 10505
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.245347373884635e-06,
      "loss": 0.6665,
      "step": 10510
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.244722513640287e-06,
      "loss": 0.6971,
      "step": 10515
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2440975517403352e-06,
      "loss": 0.6807,
      "step": 10520
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.243472488444383e-06,
      "loss": 0.6471,
      "step": 10525
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2428473240120763e-06,
      "loss": 0.658,
      "step": 10530
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2422220587031033e-06,
      "loss": 0.7214,
      "step": 10535
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2415966927771938e-06,
      "loss": 0.6445,
      "step": 10540
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2409712264941189e-06,
      "loss": 0.6899,
      "step": 10545
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.240345660113692e-06,
      "loss": 0.7066,
      "step": 10550
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2397199938957678e-06,
      "loss": 0.6361,
      "step": 10555
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2390942281002435e-06,
      "loss": 0.6871,
      "step": 10560
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.238468362987056e-06,
      "loss": 0.6544,
      "step": 10565
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.237842398816184e-06,
      "loss": 0.6198,
      "step": 10570
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.237216335847648e-06,
      "loss": 0.6907,
      "step": 10575
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.236590174341509e-06,
      "loss": 0.6411,
      "step": 10580
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2359639145578693e-06,
      "loss": 0.6724,
      "step": 10585
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2353375567568715e-06,
      "loss": 0.6606,
      "step": 10590
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.234711101198699e-06,
      "loss": 0.6745,
      "step": 10595
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2340845481435764e-06,
      "loss": 0.6417,
      "step": 10600
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.233457897851768e-06,
      "loss": 0.6824,
      "step": 10605
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2328311505835794e-06,
      "loss": 0.6726,
      "step": 10610
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2322043065993554e-06,
      "loss": 0.692,
      "step": 10615
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2315773661594817e-06,
      "loss": 0.7019,
      "step": 10620
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2309503295243844e-06,
      "loss": 0.6775,
      "step": 10625
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.230323196954528e-06,
      "loss": 0.6774,
      "step": 10630
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.229695968710419e-06,
      "loss": 0.7227,
      "step": 10635
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2290686450526021e-06,
      "loss": 0.6676,
      "step": 10640
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2284412262416621e-06,
      "loss": 0.6781,
      "step": 10645
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2278137125382235e-06,
      "loss": 0.666,
      "step": 10650
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2271861042029496e-06,
      "loss": 0.6926,
      "step": 10655
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2265584014965439e-06,
      "loss": 0.6458,
      "step": 10660
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2259306046797486e-06,
      "loss": 0.6518,
      "step": 10665
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2253027140133444e-06,
      "loss": 0.6911,
      "step": 10670
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2246747297581527e-06,
      "loss": 0.6738,
      "step": 10675
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2240466521750318e-06,
      "loss": 0.6601,
      "step": 10680
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.22341848152488e-06,
      "loss": 0.6561,
      "step": 10685
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.222790218068634e-06,
      "loss": 0.6899,
      "step": 10690
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2221618620672689e-06,
      "loss": 0.6685,
      "step": 10695
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2215334137817985e-06,
      "loss": 0.6926,
      "step": 10700
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2209048734732742e-06,
      "loss": 0.6241,
      "step": 10705
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2202762414027865e-06,
      "loss": 0.6556,
      "step": 10710
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.219647517831464e-06,
      "loss": 0.6612,
      "step": 10715
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2190187030204727e-06,
      "loss": 0.6931,
      "step": 10720
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2183897972310168e-06,
      "loss": 0.6806,
      "step": 10725
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2177608007243384e-06,
      "loss": 0.7048,
      "step": 10730
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2171317137617172e-06,
      "loss": 0.6722,
      "step": 10735
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2165025366044698e-06,
      "loss": 0.7012,
      "step": 10740
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2158732695139523e-06,
      "loss": 0.6476,
      "step": 10745
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2152439127515557e-06,
      "loss": 0.661,
      "step": 10750
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2146144665787094e-06,
      "loss": 0.6862,
      "step": 10755
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2139849312568803e-06,
      "loss": 0.6949,
      "step": 10760
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2133553070475714e-06,
      "loss": 0.6565,
      "step": 10765
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2127255942123236e-06,
      "loss": 0.6803,
      "step": 10770
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2120957930127139e-06,
      "loss": 0.6909,
      "step": 10775
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.211465903710356e-06,
      "loss": 0.6885,
      "step": 10780
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.210835926566901e-06,
      "loss": 0.6837,
      "step": 10785
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2102058618440354e-06,
      "loss": 0.676,
      "step": 10790
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2095757098034829e-06,
      "loss": 0.6624,
      "step": 10795
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2089454707070033e-06,
      "loss": 0.6218,
      "step": 10800
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.208315144816392e-06,
      "loss": 0.6528,
      "step": 10805
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2076847323934813e-06,
      "loss": 0.6933,
      "step": 10810
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2070542337001385e-06,
      "loss": 0.6646,
      "step": 10815
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.206423648998268e-06,
      "loss": 0.6935,
      "step": 10820
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.205792978549809e-06,
      "loss": 0.6853,
      "step": 10825
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2051622226167362e-06,
      "loss": 0.6648,
      "step": 10830
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2045313814610602e-06,
      "loss": 0.6587,
      "step": 10835
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.203900455344827e-06,
      "loss": 0.6662,
      "step": 10840
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2032694445301182e-06,
      "loss": 0.6504,
      "step": 10845
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2026383492790495e-06,
      "loss": 0.7228,
      "step": 10850
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2020071698537727e-06,
      "loss": 0.6674,
      "step": 10855
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2013759065164748e-06,
      "loss": 0.7009,
      "step": 10860
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.200744559529376e-06,
      "loss": 0.6588,
      "step": 10865
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2001131291547335e-06,
      "loss": 0.651,
      "step": 10870
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.1994816156548368e-06,
      "loss": 0.644,
      "step": 10875
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.1988500192920122e-06,
      "loss": 0.6669,
      "step": 10880
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.1982183403286186e-06,
      "loss": 0.6718,
      "step": 10885
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.1975865790270503e-06,
      "loss": 0.6594,
      "step": 10890
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.1969547356497356e-06,
      "loss": 0.6279,
      "step": 10895
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.1963228104591362e-06,
      "loss": 0.6866,
      "step": 10900
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.1956908037177484e-06,
      "loss": 0.6836,
      "step": 10905
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.1950587156881032e-06,
      "loss": 0.6323,
      "step": 10910
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.1944265466327634e-06,
      "loss": 0.6714,
      "step": 10915
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.193794296814327e-06,
      "loss": 0.6849,
      "step": 10920
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.193161966495425e-06,
      "loss": 0.6654,
      "step": 10925
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.1925295559387221e-06,
      "loss": 0.7011,
      "step": 10930
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.191897065406916e-06,
      "loss": 0.6452,
      "step": 10935
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.1912644951627375e-06,
      "loss": 0.6537,
      "step": 10940
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.1906318454689512e-06,
      "loss": 0.6852,
      "step": 10945
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.189999116588354e-06,
      "loss": 0.6316,
      "step": 10950
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.189366308783776e-06,
      "loss": 0.6368,
      "step": 10955
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1887334223180804e-06,
      "loss": 0.6621,
      "step": 10960
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1881004574541625e-06,
      "loss": 0.6432,
      "step": 10965
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.18746741445495e-06,
      "loss": 0.6771,
      "step": 10970
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1868342935834041e-06,
      "loss": 0.6967,
      "step": 10975
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1862010951025175e-06,
      "loss": 0.7035,
      "step": 10980
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1855678192753151e-06,
      "loss": 0.6717,
      "step": 10985
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1849344663648543e-06,
      "loss": 0.6407,
      "step": 10990
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1843010366342248e-06,
      "loss": 0.6802,
      "step": 10995
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1836675303465472e-06,
      "loss": 0.6332,
      "step": 11000
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1830339477649749e-06,
      "loss": 0.6221,
      "step": 11005
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1824002891526923e-06,
      "loss": 0.6445,
      "step": 11010
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1817665547729157e-06,
      "loss": 0.6818,
      "step": 11015
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.181132744888893e-06,
      "loss": 0.6772,
      "step": 11020
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1804988597639034e-06,
      "loss": 0.6692,
      "step": 11025
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1798648996612572e-06,
      "loss": 0.6727,
      "step": 11030
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1792308648442958e-06,
      "loss": 0.7129,
      "step": 11035
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1785967555763915e-06,
      "loss": 0.636,
      "step": 11040
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1779625721209484e-06,
      "loss": 0.6769,
      "step": 11045
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1773283147414007e-06,
      "loss": 0.663,
      "step": 11050
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1766939837012128e-06,
      "loss": 0.6988,
      "step": 11055
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1760595792638807e-06,
      "loss": 0.6399,
      "step": 11060
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1754251016929307e-06,
      "loss": 0.6709,
      "step": 11065
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.174790551251919e-06,
      "loss": 0.661,
      "step": 11070
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1741559282044322e-06,
      "loss": 0.6604,
      "step": 11075
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1735212328140876e-06,
      "loss": 0.6601,
      "step": 11080
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1728864653445314e-06,
      "loss": 0.6732,
      "step": 11085
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1722516260594412e-06,
      "loss": 0.6564,
      "step": 11090
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1716167152225237e-06,
      "loss": 0.6819,
      "step": 11095
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.170981733097515e-06,
      "loss": 0.6739,
      "step": 11100
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1703466799481817e-06,
      "loss": 0.6658,
      "step": 11105
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1697115560383185e-06,
      "loss": 0.6999,
      "step": 11110
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1690763616317507e-06,
      "loss": 0.6797,
      "step": 11115
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.168441096992333e-06,
      "loss": 0.6541,
      "step": 11120
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1678057623839484e-06,
      "loss": 0.6586,
      "step": 11125
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1671703580705094e-06,
      "loss": 0.6609,
      "step": 11130
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1665348843159574e-06,
      "loss": 0.6493,
      "step": 11135
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1658993413842624e-06,
      "loss": 0.6771,
      "step": 11140
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1652637295394244e-06,
      "loss": 0.6806,
      "step": 11145
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1646280490454696e-06,
      "loss": 0.6739,
      "step": 11150
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1639923001664555e-06,
      "loss": 0.6482,
      "step": 11155
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1633564831664656e-06,
      "loss": 0.6759,
      "step": 11160
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1627205983096135e-06,
      "loss": 0.6722,
      "step": 11165
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.16208464586004e-06,
      "loss": 0.6531,
      "step": 11170
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1614486260819138e-06,
      "loss": 0.6934,
      "step": 11175
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1608125392394327e-06,
      "loss": 0.6637,
      "step": 11180
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1601763855968212e-06,
      "loss": 0.6879,
      "step": 11185
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1595401654183322e-06,
      "loss": 0.6499,
      "step": 11190
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1589038789682456e-06,
      "loss": 0.7014,
      "step": 11195
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1582675265108699e-06,
      "loss": 0.6136,
      "step": 11200
    },
    {
      "epoch": 0.45,
      "eval_loss": 0.6401504874229431,
      "eval_runtime": 144.3337,
      "eval_samples_per_second": 16.393,
      "eval_steps_per_second": 2.737,
      "step": 11200
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.15763110831054e-06,
      "loss": 0.6646,
      "step": 11205
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1569946246316182e-06,
      "loss": 0.7085,
      "step": 11210
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.156358075738495e-06,
      "loss": 0.6621,
      "step": 11215
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1557214618955868e-06,
      "loss": 0.6703,
      "step": 11220
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1550847833673374e-06,
      "loss": 0.7204,
      "step": 11225
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.154448040418218e-06,
      "loss": 0.6923,
      "step": 11230
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1538112333127253e-06,
      "loss": 0.6608,
      "step": 11235
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1531743623153842e-06,
      "loss": 0.6824,
      "step": 11240
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1525374276907449e-06,
      "loss": 0.7322,
      "step": 11245
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1519004297033847e-06,
      "loss": 0.6432,
      "step": 11250
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1512633686179071e-06,
      "loss": 0.6795,
      "step": 11255
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1506262446989417e-06,
      "loss": 0.7229,
      "step": 11260
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.149989058211144e-06,
      "loss": 0.6954,
      "step": 11265
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.149351809419196e-06,
      "loss": 0.6879,
      "step": 11270
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.148714498587805e-06,
      "loss": 0.6642,
      "step": 11275
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1480771259817048e-06,
      "loss": 0.7015,
      "step": 11280
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.147439691865654e-06,
      "loss": 0.6467,
      "step": 11285
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1468021965044377e-06,
      "loss": 0.7045,
      "step": 11290
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1461646401628654e-06,
      "loss": 0.6635,
      "step": 11295
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1455270231057728e-06,
      "loss": 0.6943,
      "step": 11300
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.14488934559802e-06,
      "loss": 0.6626,
      "step": 11305
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1442516079044932e-06,
      "loss": 0.6716,
      "step": 11310
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1436138102901031e-06,
      "loss": 0.6748,
      "step": 11315
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.142975953019785e-06,
      "loss": 0.7028,
      "step": 11320
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1423380363584999e-06,
      "loss": 0.6044,
      "step": 11325
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1417000605712316e-06,
      "loss": 0.6831,
      "step": 11330
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1410620259229908e-06,
      "loss": 0.6632,
      "step": 11335
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1404239326788115e-06,
      "loss": 0.6393,
      "step": 11340
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1397857811037512e-06,
      "loss": 0.6501,
      "step": 11345
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1391475714628932e-06,
      "loss": 0.6398,
      "step": 11350
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.138509304021344e-06,
      "loss": 0.6784,
      "step": 11355
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1378709790442346e-06,
      "loss": 0.7065,
      "step": 11360
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1372325967967196e-06,
      "loss": 0.6189,
      "step": 11365
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1365941575439772e-06,
      "loss": 0.6652,
      "step": 11370
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1359556615512099e-06,
      "loss": 0.6752,
      "step": 11375
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1353171090836427e-06,
      "loss": 0.6668,
      "step": 11380
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.134678500406525e-06,
      "loss": 0.6587,
      "step": 11385
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.13403983578513e-06,
      "loss": 0.6873,
      "step": 11390
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1334011154847527e-06,
      "loss": 0.6975,
      "step": 11395
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1327623397707122e-06,
      "loss": 0.6784,
      "step": 11400
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1321235089083502e-06,
      "loss": 0.6643,
      "step": 11405
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1314846231630315e-06,
      "loss": 0.6754,
      "step": 11410
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1308456828001441e-06,
      "loss": 0.6689,
      "step": 11415
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1302066880850975e-06,
      "loss": 0.6594,
      "step": 11420
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1295676392833253e-06,
      "loss": 0.6416,
      "step": 11425
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1289285366602826e-06,
      "loss": 0.7223,
      "step": 11430
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1282893804814468e-06,
      "loss": 0.6944,
      "step": 11435
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.127650171012318e-06,
      "loss": 0.6598,
      "step": 11440
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1270109085184182e-06,
      "loss": 0.7102,
      "step": 11445
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1263715932652919e-06,
      "loss": 0.6803,
      "step": 11450
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1257322255185044e-06,
      "loss": 0.65,
      "step": 11455
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1250928055436443e-06,
      "loss": 0.7018,
      "step": 11460
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.12445333360632e-06,
      "loss": 0.6409,
      "step": 11465
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1238138099721634e-06,
      "loss": 0.6724,
      "step": 11470
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1231742349068271e-06,
      "loss": 0.6854,
      "step": 11475
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1225346086759846e-06,
      "loss": 0.7102,
      "step": 11480
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1218949315453314e-06,
      "loss": 0.6584,
      "step": 11485
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1212552037805836e-06,
      "loss": 0.6584,
      "step": 11490
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1206154256474786e-06,
      "loss": 0.6846,
      "step": 11495
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.119975597411775e-06,
      "loss": 0.7037,
      "step": 11500
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1193357193392512e-06,
      "loss": 0.6689,
      "step": 11505
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1186957916957078e-06,
      "loss": 0.6556,
      "step": 11510
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1180558147469642e-06,
      "loss": 0.6557,
      "step": 11515
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1174157887588623e-06,
      "loss": 0.6662,
      "step": 11520
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1167757139972626e-06,
      "loss": 0.6702,
      "step": 11525
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.116135590728047e-06,
      "loss": 0.6682,
      "step": 11530
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.115495419217117e-06,
      "loss": 0.6855,
      "step": 11535
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.114855199730394e-06,
      "loss": 0.6796,
      "step": 11540
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1142149325338199e-06,
      "loss": 0.6481,
      "step": 11545
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1135746178933563e-06,
      "loss": 0.7167,
      "step": 11550
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.112934256074984e-06,
      "loss": 0.6602,
      "step": 11555
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1122938473447038e-06,
      "loss": 0.6848,
      "step": 11560
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1116533919685361e-06,
      "loss": 0.6628,
      "step": 11565
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1110128902125201e-06,
      "loss": 0.6826,
      "step": 11570
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1103723423427153e-06,
      "loss": 0.6527,
      "step": 11575
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1097317486251992e-06,
      "loss": 0.6737,
      "step": 11580
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.109091109326069e-06,
      "loss": 0.6854,
      "step": 11585
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1084504247114406e-06,
      "loss": 0.7145,
      "step": 11590
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.107809695047449e-06,
      "loss": 0.6756,
      "step": 11595
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1071689206002474e-06,
      "loss": 0.6725,
      "step": 11600
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1065281016360083e-06,
      "loss": 0.7145,
      "step": 11605
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1058872384209224e-06,
      "loss": 0.6899,
      "step": 11610
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1052463312211983e-06,
      "loss": 0.6344,
      "step": 11615
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1046053803030637e-06,
      "loss": 0.654,
      "step": 11620
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1039643859327635e-06,
      "loss": 0.6741,
      "step": 11625
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1033233483765615e-06,
      "loss": 0.6814,
      "step": 11630
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1026822679007395e-06,
      "loss": 0.6565,
      "step": 11635
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1020411447715961e-06,
      "loss": 0.6916,
      "step": 11640
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1013999792554486e-06,
      "loss": 0.6894,
      "step": 11645
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1007587716186317e-06,
      "loss": 0.6698,
      "step": 11650
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1001175221274968e-06,
      "loss": 0.7096,
      "step": 11655
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.0994762310484142e-06,
      "loss": 0.6887,
      "step": 11660
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.0988348986477705e-06,
      "loss": 0.671,
      "step": 11665
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.0981935251919693e-06,
      "loss": 0.6727,
      "step": 11670
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.0975521109474318e-06,
      "loss": 0.6777,
      "step": 11675
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.0969106561805952e-06,
      "loss": 0.6661,
      "step": 11680
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.0962691611579154e-06,
      "loss": 0.6576,
      "step": 11685
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.0956276261458629e-06,
      "loss": 0.6415,
      "step": 11690
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0949860514109264e-06,
      "loss": 0.6485,
      "step": 11695
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.09434443721961e-06,
      "loss": 0.6966,
      "step": 11700
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0937027838384345e-06,
      "loss": 0.6746,
      "step": 11705
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.093061091533938e-06,
      "loss": 0.6565,
      "step": 11710
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0924193605726733e-06,
      "loss": 0.7127,
      "step": 11715
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0917775912212099e-06,
      "loss": 0.6738,
      "step": 11720
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0911357837461332e-06,
      "loss": 0.6712,
      "step": 11725
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0904939384140445e-06,
      "loss": 0.6617,
      "step": 11730
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0898520554915607e-06,
      "loss": 0.6919,
      "step": 11735
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0892101352453142e-06,
      "loss": 0.6731,
      "step": 11740
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0885681779419537e-06,
      "loss": 0.6931,
      "step": 11745
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0879261838481426e-06,
      "loss": 0.631,
      "step": 11750
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0872841532305587e-06,
      "loss": 0.6515,
      "step": 11755
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0866420863558969e-06,
      "loss": 0.6674,
      "step": 11760
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0859999834908657e-06,
      "loss": 0.6514,
      "step": 11765
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0853578449021896e-06,
      "loss": 0.6393,
      "step": 11770
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.084715670856607e-06,
      "loss": 0.6941,
      "step": 11775
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0840734616208712e-06,
      "loss": 0.664,
      "step": 11780
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0834312174617508e-06,
      "loss": 0.6636,
      "step": 11785
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0827889386460281e-06,
      "loss": 0.6756,
      "step": 11790
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0821466254405004e-06,
      "loss": 0.7116,
      "step": 11795
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0815042781119788e-06,
      "loss": 0.6647,
      "step": 11800
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0808618969272888e-06,
      "loss": 0.6404,
      "step": 11805
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0802194821532702e-06,
      "loss": 0.6711,
      "step": 11810
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.079577034056776e-06,
      "loss": 0.6717,
      "step": 11815
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.078934552904674e-06,
      "loss": 0.6705,
      "step": 11820
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0782920389638452e-06,
      "loss": 0.6713,
      "step": 11825
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0776494925011846e-06,
      "loss": 0.6474,
      "step": 11830
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0770069137836e-06,
      "loss": 0.6745,
      "step": 11835
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0763643030780126e-06,
      "loss": 0.6773,
      "step": 11840
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.075721660651358e-06,
      "loss": 0.6767,
      "step": 11845
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0750789867705843e-06,
      "loss": 0.6758,
      "step": 11850
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0744362817026524e-06,
      "loss": 0.686,
      "step": 11855
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0737935457145364e-06,
      "loss": 0.6736,
      "step": 11860
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.073150779073223e-06,
      "loss": 0.668,
      "step": 11865
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0725079820457123e-06,
      "loss": 0.7003,
      "step": 11870
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0718651548990163e-06,
      "loss": 0.682,
      "step": 11875
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0712222979001602e-06,
      "loss": 0.6445,
      "step": 11880
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0705794113161808e-06,
      "loss": 0.6872,
      "step": 11885
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0699364954141276e-06,
      "loss": 0.6936,
      "step": 11890
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0692935504610625e-06,
      "loss": 0.6195,
      "step": 11895
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.068650576724059e-06,
      "loss": 0.6658,
      "step": 11900
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0680075744702034e-06,
      "loss": 0.6799,
      "step": 11905
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0673645439665925e-06,
      "loss": 0.6823,
      "step": 11910
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0667214854803357e-06,
      "loss": 0.6677,
      "step": 11915
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0660783992785541e-06,
      "loss": 0.6648,
      "step": 11920
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.06543528562838e-06,
      "loss": 0.6313,
      "step": 11925
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0647921447969577e-06,
      "loss": 0.667,
      "step": 11930
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0641489770514418e-06,
      "loss": 0.6567,
      "step": 11935
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0635057826589987e-06,
      "loss": 0.6727,
      "step": 11940
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0628625618868056e-06,
      "loss": 0.6835,
      "step": 11945
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.062219315002051e-06,
      "loss": 0.6329,
      "step": 11950
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.061576042271934e-06,
      "loss": 0.6823,
      "step": 11955
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0609327439636647e-06,
      "loss": 0.6514,
      "step": 11960
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0602894203444633e-06,
      "loss": 0.716,
      "step": 11965
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0596460716815612e-06,
      "loss": 0.6778,
      "step": 11970
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.059002698242199e-06,
      "loss": 0.6671,
      "step": 11975
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0583593002936298e-06,
      "loss": 0.6936,
      "step": 11980
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0577158781031147e-06,
      "loss": 0.7211,
      "step": 11985
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0570724319379254e-06,
      "loss": 0.6667,
      "step": 11990
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0564289620653446e-06,
      "loss": 0.7181,
      "step": 11995
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0557854687526632e-06,
      "loss": 0.6431,
      "step": 12000
    },
    {
      "epoch": 0.49,
      "eval_loss": 0.6379530429840088,
      "eval_runtime": 144.8913,
      "eval_samples_per_second": 16.329,
      "eval_steps_per_second": 2.726,
      "step": 12000
    }
  ],
  "logging_steps": 5,
  "max_steps": 24619,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 400,
  "total_flos": 1676879253282816.0,
  "trial_name": null,
  "trial_params": null
}