{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9996160245744274,
  "global_step": 11718,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 0.0,
      "loss": 9.2897,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 5.555555555555556e-06,
      "loss": 9.3838,
      "step": 5
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2222222222222223e-05,
      "loss": 9.2904,
      "step": 10
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.777777777777778e-05,
      "loss": 8.364,
      "step": 15
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.555555555555556e-05,
      "loss": 7.4332,
      "step": 20
    },
    {
      "epoch": 0.01,
      "learning_rate": 7.222222222222222e-05,
      "loss": 6.7965,
      "step": 25
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001,
      "loss": 6.134,
      "step": 30
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00012777777777777776,
      "loss": 5.6268,
      "step": 35
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00015555555555555556,
      "loss": 5.3261,
      "step": 40
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00018333333333333334,
      "loss": 4.8834,
      "step": 45
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00019999998553576974,
      "loss": 4.638,
      "step": 50
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001999998228132274,
      "loss": 4.3787,
      "step": 55
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019999947928815007,
      "loss": 4.0214,
      "step": 60
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019999895496115885,
      "loss": 3.8979,
      "step": 65
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019999824983320177,
      "loss": 3.7545,
      "step": 70
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019999736390555368,
      "loss": 3.6385,
      "step": 75
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001999962971798164,
      "loss": 3.5271,
      "step": 80
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019999504965791858,
      "loss": 3.3175,
      "step": 85
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019999362134211576,
      "loss": 3.3254,
      "step": 90
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001999920122349904,
      "loss": 3.231,
      "step": 95
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019999022233945185,
      "loss": 3.1622,
      "step": 100
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0001999882516587362,
      "loss": 3.1442,
      "step": 105
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019998610019640654,
      "loss": 3.0162,
      "step": 110
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019998376795635282,
      "loss": 3.0849,
      "step": 115
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019998125494279174,
      "loss": 3.0558,
      "step": 120
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019997856116026692,
      "loss": 2.9272,
      "step": 125
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019997568661364877,
      "loss": 3.0184,
      "step": 130
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0001999726313081346,
      "loss": 2.9134,
      "step": 135
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019996939524924846,
      "loss": 2.9225,
      "step": 140
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019996597844284123,
      "loss": 2.8925,
      "step": 145
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001999623808950906,
      "loss": 2.8565,
      "step": 150
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019995860261250108,
      "loss": 2.8153,
      "step": 155
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019995464360190386,
      "loss": 2.8274,
      "step": 160
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019995050387045695,
      "loss": 2.6365,
      "step": 165
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019994618342564512,
      "loss": 2.8168,
      "step": 170
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019994168227527984,
      "loss": 2.7378,
      "step": 175
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019993700042749937,
      "loss": 2.7134,
      "step": 180
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019993213789076854,
      "loss": 2.6976,
      "step": 185
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019992709467387903,
      "loss": 2.6616,
      "step": 190
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0001999218707859491,
      "loss": 2.6653,
      "step": 195
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019991646623642364,
      "loss": 2.6252,
      "step": 200
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0001999108810350743,
      "loss": 2.6404,
      "step": 205
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019990511519199923,
      "loss": 2.7217,
      "step": 210
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019989916871762325,
      "loss": 2.6164,
      "step": 215
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0001998930416226978,
      "loss": 2.6779,
      "step": 220
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0001998867339183008,
      "loss": 2.5657,
      "step": 225
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019988024561583677,
      "loss": 2.5789,
      "step": 230
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019987357672703672,
      "loss": 2.4826,
      "step": 235
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019986672726395824,
      "loss": 2.4319,
      "step": 240
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019985969723898534,
      "loss": 2.5734,
      "step": 245
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0001998524866648285,
      "loss": 2.5765,
      "step": 250
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019984509555452461,
      "loss": 2.5343,
      "step": 255
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019983752392143704,
      "loss": 2.5778,
      "step": 260
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019982977177925552,
      "loss": 2.4845,
      "step": 265
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019982183914199616,
      "loss": 2.5486,
      "step": 270
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019981372602400138,
      "loss": 2.5074,
      "step": 275
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019980543243993992,
      "loss": 2.5488,
      "step": 280
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019979695840480682,
      "loss": 2.4273,
      "step": 285
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0001997883039339234,
      "loss": 2.4197,
      "step": 290
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019977946904293714,
      "loss": 2.4427,
      "step": 295
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0001997704537478218,
      "loss": 2.4044,
      "step": 300
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019976125806487734,
      "loss": 2.4811,
      "step": 305
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019975188201072972,
      "loss": 2.4572,
      "step": 310
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019974232560233122,
      "loss": 2.4498,
      "step": 315
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019973258885696,
      "loss": 2.5056,
      "step": 320
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019972267179222045,
      "loss": 2.4114,
      "step": 325
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019971257442604287,
      "loss": 2.4181,
      "step": 330
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019970229677668358,
      "loss": 2.3741,
      "step": 335
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0001996918388627249,
      "loss": 2.3721,
      "step": 340
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.000199681200703075,
      "loss": 2.4324,
      "step": 345
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019967038231696803,
      "loss": 2.3941,
      "step": 350
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019965938372396385,
      "loss": 2.4113,
      "step": 355
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0001996482049439483,
      "loss": 2.4253,
      "step": 360
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019963684599713291,
      "loss": 2.3272,
      "step": 365
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.000199625306904055,
      "loss": 2.4973,
      "step": 370
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019961358768557757,
      "loss": 2.2946,
      "step": 375
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019960168836288927,
      "loss": 2.4072,
      "step": 380
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019958960895750445,
      "loss": 2.393,
      "step": 385
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019957734949126304,
      "loss": 2.3143,
      "step": 390
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0001995649099863304,
      "loss": 2.382,
      "step": 395
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0001995522904651977,
      "loss": 2.3571,
      "step": 400
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0001995394909506812,
      "loss": 2.2845,
      "step": 405
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019952651146592286,
      "loss": 2.3201,
      "step": 410
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019951335203439001,
      "loss": 2.3537,
      "step": 415
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019950001267987524,
      "loss": 2.3467,
      "step": 420
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019948649342649646,
      "loss": 2.369,
      "step": 425
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019947279429869687,
      "loss": 2.342,
      "step": 430
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019945891532124495,
      "loss": 2.3362,
      "step": 435
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019944485651923418,
      "loss": 2.2557,
      "step": 440
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019943061791808338,
      "loss": 2.2849,
      "step": 445
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019941619954353627,
      "loss": 2.3154,
      "step": 450
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019940160142166172,
      "loss": 2.2588,
      "step": 455
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019938682357885357,
      "loss": 2.3105,
      "step": 460
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019937186604183053,
      "loss": 2.3508,
      "step": 465
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0001993567288376363,
      "loss": 2.2427,
      "step": 470
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019934141199363933,
      "loss": 2.2947,
      "step": 475
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019932591553753295,
      "loss": 2.2439,
      "step": 480
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019931023949733517,
      "loss": 2.2582,
      "step": 485
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019929438390138877,
      "loss": 2.2502,
      "step": 490
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019927834877836106,
      "loss": 2.3103,
      "step": 495
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019926213415724403,
      "loss": 2.2329,
      "step": 500
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019924574006735418,
      "loss": 2.2965,
      "step": 505
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019922916653833248,
      "loss": 2.3283,
      "step": 510
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0001992124136001443,
      "loss": 2.2531,
      "step": 515
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019919548128307954,
      "loss": 2.3132,
      "step": 520
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019917836961775223,
      "loss": 2.2994,
      "step": 525
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019916107863510076,
      "loss": 2.3102,
      "step": 530
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019914360836638774,
      "loss": 2.1879,
      "step": 535
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0001991259588431999,
      "loss": 2.2855,
      "step": 540
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0001991081300974481,
      "loss": 2.1905,
      "step": 545
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0001990901221613672,
      "loss": 2.2592,
      "step": 550
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019907193506751606,
      "loss": 2.289,
      "step": 555
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019905356884877747,
      "loss": 2.2288,
      "step": 560
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019903502353835813,
      "loss": 2.2417,
      "step": 565
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0001990162991697884,
      "loss": 2.2646,
      "step": 570
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00019899739577692254,
      "loss": 2.2407,
      "step": 575
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00019897831339393842,
      "loss": 2.2366,
      "step": 580
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00019895905205533746,
      "loss": 2.1751,
      "step": 585
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00019893961179594482,
      "loss": 2.282,
      "step": 590
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00019891999265090896,
      "loss": 2.2579,
      "step": 595
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00019890019465570189,
      "loss": 2.2288,
      "step": 600
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00019888021784611893,
      "loss": 2.2074,
      "step": 605
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00019886006225827875,
      "loss": 2.1922,
      "step": 610
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00019883972792862322,
      "loss": 2.2828,
      "step": 615
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00019881921489391737,
      "loss": 2.265,
      "step": 620
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0001987985231912494,
      "loss": 2.1636,
      "step": 625
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00019877765285803044,
      "loss": 2.1897,
      "step": 630
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00019875660393199466,
      "loss": 2.1403,
      "step": 635
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00019873537645119914,
      "loss": 2.0977,
      "step": 640
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00019871397045402377,
      "loss": 2.2344,
      "step": 645
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00019869238597917122,
      "loss": 2.1271,
      "step": 650
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0001986706230656668,
      "loss": 2.1469,
      "step": 655
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00019864868175285846,
      "loss": 2.1112,
      "step": 660
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0001986265620804168,
      "loss": 2.1998,
      "step": 665
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00019860426408833476,
      "loss": 2.1744,
      "step": 670
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00019858178781692774,
      "loss": 2.1173,
      "step": 675
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00019855913330683354,
      "loss": 2.1755,
      "step": 680
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0001985363005990121,
      "loss": 2.1156,
      "step": 685
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00019851328973474562,
      "loss": 2.1461,
      "step": 690
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00019849010075563843,
      "loss": 2.1229,
      "step": 695
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00019846673370361683,
      "loss": 2.1973,
      "step": 700
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00019844318862092912,
      "loss": 2.1457,
      "step": 705
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00019841946555014552,
      "loss": 2.1617,
      "step": 710
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00019839556453415797,
      "loss": 2.1935,
      "step": 715
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00019837148561618025,
      "loss": 2.1807,
      "step": 720
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00019834722883974766,
      "loss": 2.1541,
      "step": 725
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00019832279424871719,
      "loss": 2.1056,
      "step": 730
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00019829818188726728,
      "loss": 2.1242,
      "step": 735
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00019827339179989773,
      "loss": 2.1558,
      "step": 740
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00019824842403142983,
      "loss": 2.1562,
      "step": 745
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0001982232786270059,
      "loss": 2.1095,
      "step": 750
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0001981979556320896,
      "loss": 2.091,
      "step": 755
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00019817245509246568,
      "loss": 2.1305,
      "step": 760
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00019814677705423973,
      "loss": 2.0903,
      "step": 765
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00019812092156383849,
      "loss": 2.1,
      "step": 770
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00019809488866800936,
      "loss": 2.059,
      "step": 775
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00019806867841382057,
      "loss": 2.1434,
      "step": 780
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00019804229084866102,
      "loss": 2.102,
      "step": 785
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00019801572602024016,
      "loss": 2.1808,
      "step": 790
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.000197988983976588,
      "loss": 2.0807,
      "step": 795
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00019796206476605486,
      "loss": 2.1151,
      "step": 800
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00019793496843731147,
      "loss": 2.0816,
      "step": 805
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00019790769503934882,
      "loss": 2.1386,
      "step": 810
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00019788024462147788,
      "loss": 2.1359,
      "step": 815
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0001978526172333299,
      "loss": 2.1696,
      "step": 820
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00019782481292485588,
      "loss": 2.102,
      "step": 825
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00019779683174632687,
      "loss": 2.0928,
      "step": 830
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00019776867374833365,
      "loss": 2.1227,
      "step": 835
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00019774033898178667,
      "loss": 2.1159,
      "step": 840
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.000197711827497916,
      "loss": 2.0423,
      "step": 845
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00019768313934827116,
      "loss": 2.047,
      "step": 850
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00019765427458472124,
      "loss": 2.0187,
      "step": 855
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00019762523325945447,
      "loss": 2.0837,
      "step": 860
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00019759601542497848,
      "loss": 2.0592,
      "step": 865
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00019756662113411984,
      "loss": 2.0428,
      "step": 870
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00019753705044002438,
      "loss": 2.1073,
      "step": 875
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0001975073033961567,
      "loss": 2.0745,
      "step": 880
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0001974773800563003,
      "loss": 2.0325,
      "step": 885
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00019744728047455746,
      "loss": 2.0986,
      "step": 890
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00019741700470534906,
      "loss": 2.0705,
      "step": 895
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00019738655280341456,
      "loss": 2.0851,
      "step": 900
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00019735592482381188,
      "loss": 2.0603,
      "step": 905
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00019732512082191726,
      "loss": 2.1276,
      "step": 910
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00019729414085342526,
      "loss": 2.087,
      "step": 915
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0001972629849743485,
      "loss": 2.1396,
      "step": 920
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0001972316532410177,
      "loss": 2.0405,
      "step": 925
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00019720014571008158,
      "loss": 2.131,
      "step": 930
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00019716846243850663,
      "loss": 2.0722,
      "step": 935
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00019713660348357707,
      "loss": 2.1107,
      "step": 940
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00019710456890289482,
      "loss": 2.0747,
      "step": 945
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00019707235875437933,
      "loss": 2.0343,
      "step": 950
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00019703997309626747,
      "loss": 2.0575,
      "step": 955
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00019700741198711344,
      "loss": 2.0377,
      "step": 960
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00019697467548578863,
      "loss": 2.0604,
      "step": 965
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0001969417636514816,
      "loss": 2.0612,
      "step": 970
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00019690867654369786,
      "loss": 2.0221,
      "step": 975
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00019687541422225985,
      "loss": 2.0022,
      "step": 980
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00019684197674730682,
      "loss": 2.065,
      "step": 985
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00019680836417929463,
      "loss": 2.0337,
      "step": 990
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00019677457657899585,
      "loss": 2.0171,
      "step": 995
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00019674061400749935,
      "loss": 2.0215,
      "step": 1000
    },
    {
      "epoch": 0.26,
      "eval_loss": 1.8183355331420898,
      "eval_runtime": 2.0369,
      "eval_samples_per_second": 49.095,
      "eval_steps_per_second": 1.964,
      "step": 1000
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00019670647652621041,
      "loss": 1.9703,
      "step": 1005
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00019667216419685067,
      "loss": 2.0562,
      "step": 1010
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0001966376770814577,
      "loss": 2.0276,
      "step": 1015
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00019660301524238524,
      "loss": 2.0477,
      "step": 1020
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00019656817874230288,
      "loss": 2.0457,
      "step": 1025
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00019653316764419598,
      "loss": 2.0386,
      "step": 1030
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00019649798201136566,
      "loss": 1.988,
      "step": 1035
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0001964626219074285,
      "loss": 2.0161,
      "step": 1040
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0001964270873963166,
      "loss": 2.0728,
      "step": 1045
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00019639137854227738,
      "loss": 1.9877,
      "step": 1050
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00019635549540987344,
      "loss": 1.9707,
      "step": 1055
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00019631943806398252,
      "loss": 2.0202,
      "step": 1060
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00019628320656979732,
      "loss": 2.048,
      "step": 1065
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00019624680099282547,
      "loss": 1.9747,
      "step": 1070
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00019621022139888918,
      "loss": 2.0521,
      "step": 1075
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0001961734678541255,
      "loss": 1.98,
      "step": 1080
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00019613654042498587,
      "loss": 1.9254,
      "step": 1085
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00019609943917823616,
      "loss": 1.9746,
      "step": 1090
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0001960621641809564,
      "loss": 2.0459,
      "step": 1095
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00019602471550054091,
      "loss": 2.0153,
      "step": 1100
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00019598709320469798,
      "loss": 1.9552,
      "step": 1105
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00019594929736144976,
      "loss": 2.049,
      "step": 1110
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0001959113280391322,
      "loss": 2.0373,
      "step": 1115
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.000195873185306395,
      "loss": 2.034,
      "step": 1120
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00019583486923220123,
      "loss": 2.0992,
      "step": 1125
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0001957963798858275,
      "loss": 2.0166,
      "step": 1130
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00019575771733686362,
      "loss": 2.0454,
      "step": 1135
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00019571888165521254,
      "loss": 2.0252,
      "step": 1140
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00019567987291109037,
      "loss": 1.9749,
      "step": 1145
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.000195640691175026,
      "loss": 1.929,
      "step": 1150
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0001956013365178611,
      "loss": 1.9617,
      "step": 1155
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00019556180901075,
      "loss": 1.977,
      "step": 1160
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00019552210872515963,
      "loss": 1.9455,
      "step": 1165
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0001954822357328692,
      "loss": 2.0271,
      "step": 1170
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00019544219010597024,
      "loss": 1.9803,
      "step": 1175
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00019540197191686632,
      "loss": 1.9875,
      "step": 1180
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00019536158123827316,
      "loss": 1.9826,
      "step": 1185
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00019532914455203034,
      "loss": 1.9978,
      "step": 1190
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00019528844357659525,
      "loss": 2.0476,
      "step": 1195
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00019524757031693336,
      "loss": 2.0412,
      "step": 1200
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0001952065248469447,
      "loss": 2.0222,
      "step": 1205
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00019516530724084067,
      "loss": 1.9951,
      "step": 1210
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00019512391757314383,
      "loss": 1.9622,
      "step": 1215
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00019508235591868793,
      "loss": 1.995,
      "step": 1220
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00019504062235261758,
      "loss": 1.9186,
      "step": 1225
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00019499871695038826,
      "loss": 1.9743,
      "step": 1230
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0001949566397877662,
      "loss": 1.9741,
      "step": 1235
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00019491439094082807,
      "loss": 1.9279,
      "step": 1240
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00019487197048596096,
      "loss": 1.996,
      "step": 1245
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0001948293784998623,
      "loss": 1.9672,
      "step": 1250
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0001947866150595396,
      "loss": 1.9918,
      "step": 1255
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00019475228091218712,
      "loss": 1.9849,
      "step": 1260
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00019470920904930954,
      "loss": 1.9913,
      "step": 1265
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00019466596594947775,
      "loss": 1.9844,
      "step": 1270
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00019462255169087648,
      "loss": 2.0133,
      "step": 1275
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00019457896635199996,
      "loss": 1.9548,
      "step": 1280
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00019453521001165173,
      "loss": 1.9786,
      "step": 1285
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0001944912827489445,
      "loss": 1.9684,
      "step": 1290
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0001944471846433,
      "loss": 1.9318,
      "step": 1295
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00019440291577444888,
      "loss": 2.015,
      "step": 1300
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0001943584762224305,
      "loss": 1.9223,
      "step": 1305
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00019431386606759285,
      "loss": 1.9955,
      "step": 1310
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0001942690853905924,
      "loss": 1.9918,
      "step": 1315
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00019422413427239382,
      "loss": 1.9083,
      "step": 1320
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00019417901279427006,
      "loss": 2.0179,
      "step": 1325
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00019413372103780206,
      "loss": 1.9483,
      "step": 1330
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00019408825908487858,
      "loss": 1.9777,
      "step": 1335
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00019404262701769613,
      "loss": 1.8945,
      "step": 1340
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00019399682491875878,
      "loss": 1.954,
      "step": 1345
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00019395085287087809,
      "loss": 1.9467,
      "step": 1350
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00019390471095717272,
      "loss": 1.9789,
      "step": 1355
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0001938583992610687,
      "loss": 1.9677,
      "step": 1360
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00019381191786629884,
      "loss": 1.9812,
      "step": 1365
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0001937652668569028,
      "loss": 2.0083,
      "step": 1370
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00019371844631722697,
      "loss": 1.9463,
      "step": 1375
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00019367145633192423,
      "loss": 2.0202,
      "step": 1380
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00019362429698595378,
      "loss": 1.867,
      "step": 1385
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00019357696836458109,
      "loss": 1.9782,
      "step": 1390
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00019352947055337766,
      "loss": 1.9559,
      "step": 1395
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00019348180363822087,
      "loss": 1.9721,
      "step": 1400
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00019343396770529392,
      "loss": 1.961,
      "step": 1405
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00019338596284108545,
      "loss": 1.9242,
      "step": 1410
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00019333778913238972,
      "loss": 1.9001,
      "step": 1415
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0001932894466663061,
      "loss": 1.9546,
      "step": 1420
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00019324093553023917,
      "loss": 1.9429,
      "step": 1425
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00019319225581189848,
      "loss": 1.8924,
      "step": 1430
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0001931434075992983,
      "loss": 1.9765,
      "step": 1435
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00019309439098075766,
      "loss": 1.9305,
      "step": 1440
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0001930452060448999,
      "loss": 1.933,
      "step": 1445
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0001929958528806529,
      "loss": 1.9897,
      "step": 1450
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00019294633157724853,
      "loss": 1.9231,
      "step": 1455
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00019289664222422274,
      "loss": 1.9465,
      "step": 1460
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00019284678491141529,
      "loss": 1.8723,
      "step": 1465
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0001927967597289696,
      "loss": 1.9487,
      "step": 1470
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0001927465667673327,
      "loss": 1.9555,
      "step": 1475
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00019269620611725481,
      "loss": 1.9607,
      "step": 1480
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0001926456778697895,
      "loss": 1.9146,
      "step": 1485
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00019260513466307967,
      "loss": 1.9291,
      "step": 1490
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00019255430497073888,
      "loss": 1.9462,
      "step": 1495
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00019250330793757196,
      "loss": 1.8946,
      "step": 1500
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00019245214365578298,
      "loss": 1.916,
      "step": 1505
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00019241109187349817,
      "loss": 1.8882,
      "step": 1510
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00019235962677750896,
      "loss": 1.9685,
      "step": 1515
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0001923079946926774,
      "loss": 1.9601,
      "step": 1520
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0001922665688555682,
      "loss": 1.895,
      "step": 1525
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00019221463642627083,
      "loss": 1.9459,
      "step": 1530
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00019216253727027778,
      "loss": 1.9854,
      "step": 1535
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00019211027148178578,
      "loss": 1.9133,
      "step": 1540
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00019205783915529287,
      "loss": 2.0028,
      "step": 1545
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0001920052403855982,
      "loss": 1.9149,
      "step": 1550
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00019195247526780187,
      "loss": 1.8837,
      "step": 1555
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00019189954389730468,
      "loss": 1.9246,
      "step": 1560
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00019184644636980805,
      "loss": 1.9212,
      "step": 1565
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0001917931827813139,
      "loss": 1.9212,
      "step": 1570
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00019173975322812426,
      "loss": 1.9334,
      "step": 1575
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00019168615780684132,
      "loss": 1.9698,
      "step": 1580
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00019163239661436707,
      "loss": 1.9122,
      "step": 1585
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00019157846974790336,
      "loss": 1.9679,
      "step": 1590
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00019152437730495146,
      "loss": 1.9444,
      "step": 1595
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00019147011938331206,
      "loss": 1.9237,
      "step": 1600
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00019141569608108503,
      "loss": 1.9617,
      "step": 1605
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00019136110749666925,
      "loss": 1.8958,
      "step": 1610
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00019130635372876246,
      "loss": 1.8737,
      "step": 1615
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.000191251434876361,
      "loss": 1.9383,
      "step": 1620
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00019119635103875972,
      "loss": 1.8857,
      "step": 1625
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00019114110231555182,
      "loss": 1.8511,
      "step": 1630
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0001910856888066286,
      "loss": 1.911,
      "step": 1635
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00019103011061217913,
      "loss": 1.9076,
      "step": 1640
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00019097436783269047,
      "loss": 1.8789,
      "step": 1645
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00019091846056894715,
      "loss": 1.8864,
      "step": 1650
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0001908623889220311,
      "loss": 1.9381,
      "step": 1655
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00019080615299332142,
      "loss": 1.9556,
      "step": 1660
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00019074975288449432,
      "loss": 1.9341,
      "step": 1665
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00019069318869752283,
      "loss": 1.9761,
      "step": 1670
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00019063646053467652,
      "loss": 1.891,
      "step": 1675
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0001905795684985216,
      "loss": 1.9201,
      "step": 1680
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00019052251269192048,
      "loss": 1.8857,
      "step": 1685
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00019046529321803168,
      "loss": 1.9542,
      "step": 1690
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00019040791018030967,
      "loss": 1.932,
      "step": 1695
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0001903503636825046,
      "loss": 1.9582,
      "step": 1700
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00019029265382866214,
      "loss": 1.9013,
      "step": 1705
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00019023478072312342,
      "loss": 1.9272,
      "step": 1710
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00019017674447052463,
      "loss": 1.8687,
      "step": 1715
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.000190118545175797,
      "loss": 1.9503,
      "step": 1720
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00019006018294416647,
      "loss": 1.8982,
      "step": 1725
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0001900016578811537,
      "loss": 1.9493,
      "step": 1730
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0001899429700925736,
      "loss": 1.8548,
      "step": 1735
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00018989590277059657,
      "loss": 1.8518,
      "step": 1740
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0001898723300979511,
      "loss": 1.9131,
      "step": 1745
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00018981328468704127,
      "loss": 1.8887,
      "step": 1750
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00018975407689114825,
      "loss": 1.8728,
      "step": 1755
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00018969470681732143,
      "loss": 1.9128,
      "step": 1760
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00018963517457290361,
      "loss": 1.9224,
      "step": 1765
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00018957548026553073,
      "loss": 1.8755,
      "step": 1770
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0001895156240031319,
      "loss": 1.9126,
      "step": 1775
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00018945560589392888,
      "loss": 1.9315,
      "step": 1780
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0001893954260464361,
      "loss": 1.8778,
      "step": 1785
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00018933508456946054,
      "loss": 1.8613,
      "step": 1790
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00018927458157210123,
      "loss": 1.8284,
      "step": 1795
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00018921391716374933,
      "loss": 1.8996,
      "step": 1800
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00018915309145408786,
      "loss": 1.9148,
      "step": 1805
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0001890921045530914,
      "loss": 1.8774,
      "step": 1810
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00018903095657102603,
      "loss": 1.8948,
      "step": 1815
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00018896964761844903,
      "loss": 1.8544,
      "step": 1820
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00018890817780620878,
      "loss": 1.7928,
      "step": 1825
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0001888465472454444,
      "loss": 1.9044,
      "step": 1830
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00018878475604758573,
      "loss": 1.9089,
      "step": 1835
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00018872280432435302,
      "loss": 1.8255,
      "step": 1840
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00018866069218775676,
      "loss": 1.9163,
      "step": 1845
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00018859841975009745,
      "loss": 1.8495,
      "step": 1850
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00018853598712396545,
      "loss": 1.8004,
      "step": 1855
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00018847339442224073,
      "loss": 1.9387,
      "step": 1860
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00018841064175809268,
      "loss": 1.8566,
      "step": 1865
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0001883477292449799,
      "loss": 1.8849,
      "step": 1870
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00018828465699665,
      "loss": 1.903,
      "step": 1875
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00018822142512713944,
      "loss": 1.8851,
      "step": 1880
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0001881580337507732,
      "loss": 1.8783,
      "step": 1885
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0001880944829821648,
      "loss": 1.8592,
      "step": 1890
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00018803077293621568,
      "loss": 1.8978,
      "step": 1895
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00018796690372811556,
      "loss": 1.8318,
      "step": 1900
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00018790287547334176,
      "loss": 1.8377,
      "step": 1905
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00018783868828765914,
      "loss": 1.8675,
      "step": 1910
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00018777434228712002,
      "loss": 1.8989,
      "step": 1915
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00018770983758806378,
      "loss": 1.8781,
      "step": 1920
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00018764517430711676,
      "loss": 1.8795,
      "step": 1925
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00018758035256119204,
      "loss": 1.8826,
      "step": 1930
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00018751537246748917,
      "loss": 1.9236,
      "step": 1935
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.000187450234143494,
      "loss": 1.8728,
      "step": 1940
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00018738493770697852,
      "loss": 1.8633,
      "step": 1945
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0001873194832760005,
      "loss": 1.876,
      "step": 1950
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0001872538709689035,
      "loss": 1.8794,
      "step": 1955
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0001871881009043163,
      "loss": 1.8341,
      "step": 1960
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00018712217320115322,
      "loss": 1.857,
      "step": 1965
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00018705608797861333,
      "loss": 1.9111,
      "step": 1970
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00018698984535618065,
      "loss": 1.8776,
      "step": 1975
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0001869234454536237,
      "loss": 1.909,
      "step": 1980
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0001868568883909954,
      "loss": 1.9096,
      "step": 1985
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00018679017428863283,
      "loss": 1.8574,
      "step": 1990
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00018672330326715696,
      "loss": 1.8551,
      "step": 1995
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0001866562754474726,
      "loss": 1.8565,
      "step": 2000
    },
    {
      "epoch": 0.51,
      "eval_loss": 1.6405518054962158,
      "eval_runtime": 2.0249,
      "eval_samples_per_second": 49.384,
      "eval_steps_per_second": 1.975,
      "step": 2000
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00018658909095076783,
      "loss": 1.8241,
      "step": 2005
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00018652174989851428,
      "loss": 1.9028,
      "step": 2010
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00018645425241246638,
      "loss": 1.9159,
      "step": 2015
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00018638659861466156,
      "loss": 1.8554,
      "step": 2020
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0001863187886274198,
      "loss": 1.934,
      "step": 2025
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00018625082257334353,
      "loss": 1.9281,
      "step": 2030
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00018618270057531728,
      "loss": 1.8402,
      "step": 2035
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00018611442275650756,
      "loss": 1.8373,
      "step": 2040
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00018604598924036267,
      "loss": 1.8416,
      "step": 2045
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0001859774001506123,
      "loss": 1.7969,
      "step": 2050
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00018590865561126757,
      "loss": 1.8583,
      "step": 2055
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00018583975574662046,
      "loss": 1.9004,
      "step": 2060
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0001857707006812439,
      "loss": 1.8759,
      "step": 2065
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00018570149053999152,
      "loss": 1.8812,
      "step": 2070
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00018563212544799718,
      "loss": 1.8436,
      "step": 2075
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0001855626055306749,
      "loss": 1.876,
      "step": 2080
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00018549293091371874,
      "loss": 1.9006,
      "step": 2085
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00018542310172310237,
      "loss": 1.8599,
      "step": 2090
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00018535311808507899,
      "loss": 1.8957,
      "step": 2095
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00018528298012618095,
      "loss": 1.8305,
      "step": 2100
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00018521268797321974,
      "loss": 1.7519,
      "step": 2105
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00018514224175328553,
      "loss": 1.8212,
      "step": 2110
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00018507164159374718,
      "loss": 1.859,
      "step": 2115
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0001850008876222517,
      "loss": 1.8488,
      "step": 2120
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00018492997996672438,
      "loss": 1.8126,
      "step": 2125
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0001848589187553682,
      "loss": 1.9146,
      "step": 2130
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0001847877041166639,
      "loss": 1.8637,
      "step": 2135
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.00018471633617936958,
      "loss": 1.8365,
      "step": 2140
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.00018464481507252055,
      "loss": 1.8485,
      "step": 2145
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0001845731409254289,
      "loss": 1.8374,
      "step": 2150
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.000184530063031755,
      "loss": 1.8241,
      "step": 2155
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.00018445814428993525,
      "loss": 1.8591,
      "step": 2160
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.00018438607284537907,
      "loss": 1.8777,
      "step": 2165
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00018431384882839364,
      "loss": 1.8157,
      "step": 2170
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0001842414723695621,
      "loss": 1.8832,
      "step": 2175
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00018416894359974312,
      "loss": 1.826,
      "step": 2180
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00018409626265007085,
      "loss": 1.8825,
      "step": 2185
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00018402342965195443,
      "loss": 1.896,
      "step": 2190
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00018395044473707803,
      "loss": 1.8501,
      "step": 2195
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0001838773080374005,
      "loss": 1.8059,
      "step": 2200
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00018380401968515507,
      "loss": 1.8274,
      "step": 2205
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00018373057981284913,
      "loss": 1.8822,
      "step": 2210
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0001836569885532641,
      "loss": 1.9251,
      "step": 2215
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00018358324603945508,
      "loss": 1.8399,
      "step": 2220
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00018350935240475064,
      "loss": 1.8371,
      "step": 2225
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00018343530778275254,
      "loss": 1.7737,
      "step": 2230
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00018336111230733563,
      "loss": 1.8708,
      "step": 2235
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00018328676611264745,
      "loss": 1.8698,
      "step": 2240
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00018322718072934268,
      "loss": 1.9078,
      "step": 2245
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00018315256357888762,
      "loss": 1.8251,
      "step": 2250
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00018307779608622333,
      "loss": 1.8327,
      "step": 2255
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00018300287838653154,
      "loss": 1.8461,
      "step": 2260
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0001829278106152656,
      "loss": 1.8323,
      "step": 2265
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00018285259290815023,
      "loss": 1.8486,
      "step": 2270
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00018277722540118116,
      "loss": 1.8485,
      "step": 2275
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00018270170823062498,
      "loss": 1.8796,
      "step": 2280
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0001826260415330189,
      "loss": 1.8386,
      "step": 2285
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0001825502254451705,
      "loss": 1.849,
      "step": 2290
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00018247426010415736,
      "loss": 1.7897,
      "step": 2295
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00018239814564732706,
      "loss": 1.8165,
      "step": 2300
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00018232188221229664,
      "loss": 1.925,
      "step": 2305
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00018224546993695265,
      "loss": 1.8233,
      "step": 2310
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00018216890895945062,
      "loss": 1.8417,
      "step": 2315
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00018209219941821498,
      "loss": 1.8409,
      "step": 2320
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0001820153414519388,
      "loss": 1.7978,
      "step": 2325
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00018193833519958343,
      "loss": 1.8343,
      "step": 2330
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00018186118080037842,
      "loss": 1.8589,
      "step": 2335
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0001817838783938211,
      "loss": 1.8368,
      "step": 2340
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00018170642811967644,
      "loss": 1.8111,
      "step": 2345
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00018162883011797672,
      "loss": 1.7967,
      "step": 2350
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00018155108452902138,
      "loss": 1.8485,
      "step": 2355
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00018147319149337666,
      "loss": 1.8424,
      "step": 2360
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00018139515115187535,
      "loss": 1.8031,
      "step": 2365
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00018131696364561667,
      "loss": 1.8247,
      "step": 2370
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00018123862911596588,
      "loss": 1.8358,
      "step": 2375
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0001811915578858546,
      "loss": 1.8729,
      "step": 2380
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0001811444338089425,
      "loss": 1.8388,
      "step": 2385
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00018106577632674418,
      "loss": 1.827,
      "step": 2390
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00018098697227530755,
      "loss": 1.8459,
      "step": 2395
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00018090802179711264,
      "loss": 1.824,
      "step": 2400
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0001808289250349041,
      "loss": 1.8591,
      "step": 2405
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00018074968213169115,
      "loss": 1.8501,
      "step": 2410
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00018067029323074724,
      "loss": 1.8179,
      "step": 2415
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00018059075847560974,
      "loss": 1.8307,
      "step": 2420
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0001805110780100798,
      "loss": 1.8345,
      "step": 2425
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00018043125197822192,
      "loss": 1.8327,
      "step": 2430
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0001803512805243639,
      "loss": 1.8299,
      "step": 2435
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00018027116379309638,
      "loss": 1.8365,
      "step": 2440
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00018019090192927272,
      "loss": 1.8405,
      "step": 2445
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00018011049507800863,
      "loss": 1.7728,
      "step": 2450
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.000180029943384682,
      "loss": 1.8139,
      "step": 2455
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0001799492469949326,
      "loss": 1.8555,
      "step": 2460
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00017986840605466187,
      "loss": 1.7641,
      "step": 2465
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00017978742071003242,
      "loss": 1.867,
      "step": 2470
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00017970629110746815,
      "loss": 1.8666,
      "step": 2475
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00017962501739365364,
      "loss": 1.7786,
      "step": 2480
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00017954359971553415,
      "loss": 1.7452,
      "step": 2485
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00017946203822031512,
      "loss": 1.7882,
      "step": 2490
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00017938033305546207,
      "loss": 1.8475,
      "step": 2495
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00017929848436870028,
      "loss": 1.8235,
      "step": 2500
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0001792164923080145,
      "loss": 1.8675,
      "step": 2505
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00017913435702164875,
      "loss": 1.836,
      "step": 2510
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00017905207865810593,
      "loss": 1.7496,
      "step": 2515
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0001789696573661477,
      "loss": 1.8329,
      "step": 2520
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0001788870932947941,
      "loss": 1.7659,
      "step": 2525
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00017880438659332332,
      "loss": 1.8404,
      "step": 2530
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00017872153741127145,
      "loss": 1.7743,
      "step": 2535
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00017863854589843215,
      "loss": 1.8238,
      "step": 2540
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00017855541220485648,
      "loss": 1.8287,
      "step": 2545
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00017847213648085255,
      "loss": 1.7769,
      "step": 2550
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00017838871887698518,
      "loss": 1.8508,
      "step": 2555
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0001783051595440758,
      "loss": 1.8107,
      "step": 2560
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00017822145863320214,
      "loss": 1.8357,
      "step": 2565
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00017813761629569772,
      "loss": 1.8268,
      "step": 2570
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00017805363268315197,
      "loss": 1.8067,
      "step": 2575
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00017796950794740965,
      "loss": 1.8515,
      "step": 2580
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00017788524224057067,
      "loss": 1.7839,
      "step": 2585
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0001778008357149898,
      "loss": 1.864,
      "step": 2590
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00017771628852327647,
      "loss": 1.8346,
      "step": 2595
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00017763160081829446,
      "loss": 1.7779,
      "step": 2600
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0001775467727531615,
      "loss": 1.8834,
      "step": 2605
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00017746180448124918,
      "loss": 1.7606,
      "step": 2610
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00017737669615618256,
      "loss": 1.7948,
      "step": 2615
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00017729144793183992,
      "loss": 1.778,
      "step": 2620
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00017720605996235248,
      "loss": 1.8107,
      "step": 2625
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00017712053240210413,
      "loss": 1.7959,
      "step": 2630
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00017703486540573117,
      "loss": 1.7941,
      "step": 2635
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00017694905912812192,
      "loss": 1.8386,
      "step": 2640
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0001768631137244166,
      "loss": 1.7738,
      "step": 2645
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00017679425733507803,
      "loss": 1.755,
      "step": 2650
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00017670806189615585,
      "loss": 1.8204,
      "step": 2655
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0001766217277668675,
      "loss": 1.8402,
      "step": 2660
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00017653525510330755,
      "loss": 1.8475,
      "step": 2665
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00017644864406182102,
      "loss": 1.8359,
      "step": 2670
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00017636189479900316,
      "loss": 1.7782,
      "step": 2675
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00017627500747169908,
      "loss": 1.7502,
      "step": 2680
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00017618798223700363,
      "loss": 1.776,
      "step": 2685
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00017611826286165232,
      "loss": 1.8608,
      "step": 2690
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00017603098979032683,
      "loss": 1.7792,
      "step": 2695
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0001759435792528008,
      "loss": 1.7932,
      "step": 2700
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0001758560314071151,
      "loss": 1.8069,
      "step": 2705
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00017576834641155865,
      "loss": 1.8374,
      "step": 2710
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00017568052442466843,
      "loss": 1.7983,
      "step": 2715
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00017559256560522912,
      "loss": 1.7921,
      "step": 2720
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00017550447011227278,
      "loss": 1.8499,
      "step": 2725
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00017541623810507856,
      "loss": 1.8161,
      "step": 2730
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00017532786974317244,
      "loss": 1.8425,
      "step": 2735
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00017523936518632697,
      "loss": 1.8023,
      "step": 2740
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0001751507245945609,
      "loss": 1.8118,
      "step": 2745
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00017506194812813897,
      "loss": 1.8096,
      "step": 2750
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00017497303594757153,
      "loss": 1.8208,
      "step": 2755
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00017488398821361435,
      "loss": 1.8271,
      "step": 2760
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00017479480508726832,
      "loss": 1.7719,
      "step": 2765
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00017470548672977903,
      "loss": 1.8406,
      "step": 2770
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00017461603330263663,
      "loss": 1.8597,
      "step": 2775
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0001745264449675755,
      "loss": 1.7879,
      "step": 2780
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00017443672188657386,
      "loss": 1.7856,
      "step": 2785
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0001743468642218536,
      "loss": 1.7796,
      "step": 2790
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00017425687213588003,
      "loss": 1.7924,
      "step": 2795
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0001741667457913613,
      "loss": 1.7707,
      "step": 2800
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0001740764853512485,
      "loss": 1.8271,
      "step": 2805
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00017398609097873501,
      "loss": 1.7561,
      "step": 2810
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00017389556283725654,
      "loss": 1.7428,
      "step": 2815
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00017380490109049042,
      "loss": 1.8217,
      "step": 2820
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00017371410590235578,
      "loss": 1.7255,
      "step": 2825
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00017362317743701292,
      "loss": 1.8414,
      "step": 2830
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.000173532115858863,
      "loss": 1.8338,
      "step": 2835
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00017344092133254803,
      "loss": 1.782,
      "step": 2840
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0001733495940229503,
      "loss": 1.839,
      "step": 2845
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0001732581340951922,
      "loss": 1.8006,
      "step": 2850
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0001731665417146359,
      "loss": 1.8208,
      "step": 2855
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.000173074817046883,
      "loss": 1.8102,
      "step": 2860
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00017298296025777433,
      "loss": 1.7616,
      "step": 2865
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00017289097151338965,
      "loss": 1.8392,
      "step": 2870
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00017279885098004717,
      "loss": 1.7347,
      "step": 2875
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00017270659882430347,
      "loss": 1.7882,
      "step": 2880
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00017261421521295306,
      "loss": 1.7935,
      "step": 2885
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00017252170031302816,
      "loss": 1.7692,
      "step": 2890
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00017242905429179834,
      "loss": 1.8289,
      "step": 2895
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00017233627731677028,
      "loss": 1.7925,
      "step": 2900
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0001722433695556874,
      "loss": 1.774,
      "step": 2905
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00017215033117652954,
      "loss": 1.8013,
      "step": 2910
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0001720571623475128,
      "loss": 1.8238,
      "step": 2915
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.000171963863237089,
      "loss": 1.775,
      "step": 2920
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00017187043401394573,
      "loss": 1.8423,
      "step": 2925
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00017179559706778118,
      "loss": 1.7738,
      "step": 2930
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0001717019340675858,
      "loss": 1.8124,
      "step": 2935
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00017160814142824627,
      "loss": 1.775,
      "step": 2940
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00017151421931934242,
      "loss": 1.7789,
      "step": 2945
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00017142016791068802,
      "loss": 1.7481,
      "step": 2950
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0001713259873723308,
      "loss": 1.7979,
      "step": 2955
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00017123167787455175,
      "loss": 1.7033,
      "step": 2960
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00017113723958786525,
      "loss": 1.8301,
      "step": 2965
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00017104267268301836,
      "loss": 1.7979,
      "step": 2970
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00017094797733099073,
      "loss": 1.7681,
      "step": 2975
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00017085315370299426,
      "loss": 1.8129,
      "step": 2980
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0001707582019704728,
      "loss": 1.7753,
      "step": 2985
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00017066312230510183,
      "loss": 1.7478,
      "step": 2990
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00017056791487878804,
      "loss": 1.7916,
      "step": 2995
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00017047257986366917,
      "loss": 1.7704,
      "step": 3000
    },
    {
      "epoch": 0.77,
      "eval_loss": 1.5884343385696411,
      "eval_runtime": 2.03,
      "eval_samples_per_second": 49.262,
      "eval_steps_per_second": 1.97,
      "step": 3000
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0001703771174321137,
      "loss": 1.7495,
      "step": 3005
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00017028152775672042,
      "loss": 1.8345,
      "step": 3010
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00017018581101031814,
      "loss": 1.7485,
      "step": 3015
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00017008996736596552,
      "loss": 1.8157,
      "step": 3020
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0001699939969969506,
      "loss": 1.8166,
      "step": 3025
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00016989790007679054,
      "loss": 1.7762,
      "step": 3030
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0001698209315405876,
      "loss": 1.8121,
      "step": 3035
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00016974388221190808,
      "loss": 1.833,
      "step": 3040
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00016964745707247203,
      "loss": 1.7162,
      "step": 3045
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00016955090600844492,
      "loss": 1.7007,
      "step": 3050
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00016945422919439383,
      "loss": 1.736,
      "step": 3055
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0001693574268051132,
      "loss": 1.7525,
      "step": 3060
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00016926049901562452,
      "loss": 1.7361,
      "step": 3065
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.000169163446001176,
      "loss": 1.7872,
      "step": 3070
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.00016906626793724224,
      "loss": 1.8129,
      "step": 3075
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.00016896896499952403,
      "loss": 1.8149,
      "step": 3080
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.00016887153736394774,
      "loss": 1.7877,
      "step": 3085
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.00016877398520666537,
      "loss": 1.7853,
      "step": 3090
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.00016867630870405406,
      "loss": 1.7835,
      "step": 3095
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.00016857850803271565,
      "loss": 1.7098,
      "step": 3100
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.00016848058336947657,
      "loss": 1.7589,
      "step": 3105
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0001683825348913874,
      "loss": 1.767,
      "step": 3110
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0001682843627757226,
      "loss": 1.8099,
      "step": 3115
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00016818606719998014,
      "loss": 1.7039,
      "step": 3120
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00016808764834188122,
      "loss": 1.7686,
      "step": 3125
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00016798910637937,
      "loss": 1.7712,
      "step": 3130
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0001678904414906131,
      "loss": 1.7679,
      "step": 3135
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00016779165385399952,
      "loss": 1.7529,
      "step": 3140
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00016769274364814005,
      "loss": 1.7715,
      "step": 3145
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00016759371105186719,
      "loss": 1.7047,
      "step": 3150
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00016749455624423475,
      "loss": 1.8295,
      "step": 3155
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0001673952794045174,
      "loss": 1.751,
      "step": 3160
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00016729588071221055,
      "loss": 1.7657,
      "step": 3165
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00016719636034702982,
      "loss": 1.7863,
      "step": 3170
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0001670967184889109,
      "loss": 1.7807,
      "step": 3175
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00016699695531800912,
      "loss": 1.7969,
      "step": 3180
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00016691705755728843,
      "loss": 1.7769,
      "step": 3185
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00016681707647806884,
      "loss": 1.8018,
      "step": 3190
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0001667169745916671,
      "loss": 1.7613,
      "step": 3195
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0001666167520790703,
      "loss": 1.7569,
      "step": 3200
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00016651640912148362,
      "loss": 1.7482,
      "step": 3205
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00016641594590032993,
      "loss": 1.7362,
      "step": 3210
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0001663153625972497,
      "loss": 1.7373,
      "step": 3215
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00016621465939410033,
      "loss": 1.7994,
      "step": 3220
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00016611383647295615,
      "loss": 1.8034,
      "step": 3225
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00016601289401610786,
      "loss": 1.771,
      "step": 3230
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00016591183220606226,
      "loss": 1.7844,
      "step": 3235
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00016581065122554204,
      "loss": 1.7604,
      "step": 3240
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0001657093512574853,
      "loss": 1.7697,
      "step": 3245
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00016562822573508532,
      "loss": 1.7499,
      "step": 3250
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00016552671205115335,
      "loss": 1.7472,
      "step": 3255
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00016542507989305492,
      "loss": 1.7398,
      "step": 3260
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00016532332944454395,
      "loss": 1.7206,
      "step": 3265
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00016522146088958817,
      "loss": 1.8041,
      "step": 3270
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00016511947441236877,
      "loss": 1.7728,
      "step": 3275
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00016501737019728027,
      "loss": 1.7138,
      "step": 3280
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00016491514842893,
      "loss": 1.7569,
      "step": 3285
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00016481280929213777,
      "loss": 1.7633,
      "step": 3290
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00016471035297193576,
      "loss": 1.73,
      "step": 3295
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0001646283036681959,
      "loss": 1.7086,
      "step": 3300
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00016452563688481312,
      "loss": 1.7338,
      "step": 3305
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0001644228534372364,
      "loss": 1.7476,
      "step": 3310
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00016431995351130118,
      "loss": 1.7882,
      "step": 3315
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00016421693729305346,
      "loss": 1.7481,
      "step": 3320
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0001641138049687495,
      "loss": 1.7341,
      "step": 3325
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0001640105567248555,
      "loss": 1.7624,
      "step": 3330
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0001639071927480472,
      "loss": 1.6861,
      "step": 3335
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00016380371322520968,
      "loss": 1.7322,
      "step": 3340
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0001637001183434368,
      "loss": 1.7089,
      "step": 3345
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0001635964082900311,
      "loss": 1.7919,
      "step": 3350
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00016349258325250334,
      "loss": 1.7282,
      "step": 3355
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00016338864341857213,
      "loss": 1.7253,
      "step": 3360
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00016328458897616366,
      "loss": 1.7694,
      "step": 3365
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0001631804201134113,
      "loss": 1.7304,
      "step": 3370
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0001630761370186554,
      "loss": 1.7418,
      "step": 3375
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0001629717398804427,
      "loss": 1.7013,
      "step": 3380
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.00016286722888752623,
      "loss": 1.7637,
      "step": 3385
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.00016276260422886488,
      "loss": 1.8182,
      "step": 3390
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.00016265786609362298,
      "loss": 1.7593,
      "step": 3395
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0001625530146711701,
      "loss": 1.6767,
      "step": 3400
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0001624480501510806,
      "loss": 1.7853,
      "step": 3405
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.00016234297272313332,
      "loss": 1.7319,
      "step": 3410
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.00016223778257731125,
      "loss": 1.7088,
      "step": 3415
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.00016213247990380116,
      "loss": 1.7661,
      "step": 3420
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0001620270648929933,
      "loss": 1.7962,
      "step": 3425
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0001619215377354811,
      "loss": 1.7268,
      "step": 3430
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.00016181589862206052,
      "loss": 1.7392,
      "step": 3435
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.00016171014774373022,
      "loss": 1.7416,
      "step": 3440
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.00016160428529169075,
      "loss": 1.7915,
      "step": 3445
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0001615407143447017,
      "loss": 1.7038,
      "step": 3450
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0001614346737729282,
      "loss": 1.7548,
      "step": 3455
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0001613285221255104,
      "loss": 1.7841,
      "step": 3460
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00016122225959437346,
      "loss": 1.768,
      "step": 3465
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.000161115886371643,
      "loss": 1.7726,
      "step": 3470
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.000161009402649645,
      "loss": 1.7424,
      "step": 3475
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00016090280862090495,
      "loss": 1.7308,
      "step": 3480
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0001607961044781479,
      "loss": 1.7352,
      "step": 3485
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00016068929041429804,
      "loss": 1.7744,
      "step": 3490
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00016058236662247824,
      "loss": 1.7397,
      "step": 3495
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00016047533329600975,
      "loss": 1.7218,
      "step": 3500
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00016036819062841188,
      "loss": 1.7494,
      "step": 3505
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00016026093881340163,
      "loss": 1.7849,
      "step": 3510
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00016015357804489337,
      "loss": 1.7749,
      "step": 3515
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00016004610851699837,
      "loss": 1.8122,
      "step": 3520
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0001599385304240247,
      "loss": 1.7413,
      "step": 3525
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00015983084396047653,
      "loss": 1.7451,
      "step": 3530
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00015972304932105412,
      "loss": 1.7287,
      "step": 3535
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.00015961514670065325,
      "loss": 1.7468,
      "step": 3540
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.00015950713629436493,
      "loss": 1.7306,
      "step": 3545
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.00015939901829747506,
      "loss": 1.7094,
      "step": 3550
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.00015929079290546408,
      "loss": 1.693,
      "step": 3555
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.00015918246031400654,
      "loss": 1.7384,
      "step": 3560
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.00015907402071897088,
      "loss": 1.7206,
      "step": 3565
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.00015896547431641902,
      "loss": 1.7526,
      "step": 3570
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00015885682130260592,
      "loss": 1.7292,
      "step": 3575
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00015874806187397936,
      "loss": 1.6798,
      "step": 3580
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00015863919622717947,
      "loss": 1.7443,
      "step": 3585
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00015853022455903845,
      "loss": 1.7708,
      "step": 3590
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00015842114706658017,
      "loss": 1.7308,
      "step": 3595
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00015831196394701987,
      "loss": 1.725,
      "step": 3600
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00015820267539776376,
      "loss": 1.7344,
      "step": 3605
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00015809328161640865,
      "loss": 1.7279,
      "step": 3610
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00015798378280074165,
      "loss": 1.7488,
      "step": 3615
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00015787417914873967,
      "loss": 1.7,
      "step": 3620
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00015776447085856927,
      "loss": 1.7771,
      "step": 3625
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0001576546581285862,
      "loss": 1.771,
      "step": 3630
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00015754474115733502,
      "loss": 1.693,
      "step": 3635
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00015743472014354869,
      "loss": 1.7527,
      "step": 3640
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00015732459528614842,
      "loss": 1.731,
      "step": 3645
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00015721436678424304,
      "loss": 1.7262,
      "step": 3650
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00015710403483712885,
      "loss": 1.765,
      "step": 3655
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00015699359964428916,
      "loss": 1.7769,
      "step": 3660
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00015688306140539395,
      "loss": 1.8061,
      "step": 3665
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00015677242032029952,
      "loss": 1.7197,
      "step": 3670
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0001566616765890481,
      "loss": 1.746,
      "step": 3675
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00015655083041186754,
      "loss": 1.7236,
      "step": 3680
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00015643988198917085,
      "loss": 1.7476,
      "step": 3685
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00015632883152155597,
      "loss": 1.6906,
      "step": 3690
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00015621767920980534,
      "loss": 1.7522,
      "step": 3695
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0001561064252548855,
      "loss": 1.7211,
      "step": 3700
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00015599506985794674,
      "loss": 1.7704,
      "step": 3705
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00015588361322032283,
      "loss": 1.7486,
      "step": 3710
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00015577205554353056,
      "loss": 1.6846,
      "step": 3715
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0001556603970292694,
      "loss": 1.7253,
      "step": 3720
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0001555709977505413,
      "loss": 1.6592,
      "step": 3725
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00015545915823770047,
      "loss": 1.7007,
      "step": 3730
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00015534721845311797,
      "loss": 1.695,
      "step": 3735
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0001552351785991842,
      "loss": 1.7151,
      "step": 3740
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00015512303887847036,
      "loss": 1.7295,
      "step": 3745
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00015501079949372824,
      "loss": 1.7069,
      "step": 3750
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00015489846064788994,
      "loss": 1.7176,
      "step": 3755
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0001547860225440672,
      "loss": 1.6711,
      "step": 3760
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00015467348538555141,
      "loss": 1.73,
      "step": 3765
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0001545608493758129,
      "loss": 1.7381,
      "step": 3770
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00015444811471850085,
      "loss": 1.7319,
      "step": 3775
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00015433528161744274,
      "loss": 1.7279,
      "step": 3780
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00015422235027664393,
      "loss": 1.7489,
      "step": 3785
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00015410932090028764,
      "loss": 1.7643,
      "step": 3790
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00015401882695092537,
      "loss": 1.7515,
      "step": 3795
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00015392827049430752,
      "loss": 1.7379,
      "step": 3800
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00015381498718234938,
      "loss": 1.6704,
      "step": 3805
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00015370160657135782,
      "loss": 1.7248,
      "step": 3810
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0001535881288663282,
      "loss": 1.7526,
      "step": 3815
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00015347455427243146,
      "loss": 1.7318,
      "step": 3820
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00015336088299501373,
      "loss": 1.7539,
      "step": 3825
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00015324711523959588,
      "loss": 1.7613,
      "step": 3830
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0001531332512118733,
      "loss": 1.7455,
      "step": 3835
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00015301929111771542,
      "loss": 1.7224,
      "step": 3840
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00015290523516316528,
      "loss": 1.818,
      "step": 3845
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0001527910835544394,
      "loss": 1.7811,
      "step": 3850
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00015267683649792702,
      "loss": 1.7031,
      "step": 3855
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0001525624942001902,
      "loss": 1.7497,
      "step": 3860
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00015244805686796302,
      "loss": 1.6636,
      "step": 3865
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00015233352470815143,
      "loss": 1.7377,
      "step": 3870
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0001522188979278329,
      "loss": 1.7465,
      "step": 3875
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00015210417673425583,
      "loss": 1.7301,
      "step": 3880
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00015198936133483952,
      "loss": 1.6354,
      "step": 3885
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00015187445193717342,
      "loss": 1.7327,
      "step": 3890
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00015175944874901704,
      "loss": 1.7322,
      "step": 3895
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00015164435197829944,
      "loss": 1.7583,
      "step": 3900
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0001515291618331188,
      "loss": 1.7074,
      "step": 3905
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00015146000301290025,
      "loss": 1.5754,
      "step": 3910
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0001513446639018444,
      "loss": 1.474,
      "step": 3915
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0001512292319581702,
      "loss": 1.4232,
      "step": 3920
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00015111370739058182,
      "loss": 1.4191,
      "step": 3925
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.00015099809040795102,
      "loss": 1.3778,
      "step": 3930
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.00015088238121931658,
      "loss": 1.461,
      "step": 3935
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.000150766580033884,
      "loss": 1.4005,
      "step": 3940
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0001506506870610252,
      "loss": 1.4257,
      "step": 3945
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.00015053470251027794,
      "loss": 1.3933,
      "step": 3950
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.00015041862659134558,
      "loss": 1.413,
      "step": 3955
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.00015030245951409675,
      "loss": 1.4571,
      "step": 3960
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.00015018620148856478,
      "loss": 1.3696,
      "step": 3965
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00015006985272494755,
      "loss": 1.4168,
      "step": 3970
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00014995341343360694,
      "loss": 1.3433,
      "step": 3975
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00014983688382506848,
      "loss": 1.4284,
      "step": 3980
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00014972026411002107,
      "loss": 1.4177,
      "step": 3985
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00014960355449931644,
      "loss": 1.4491,
      "step": 3990
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00014948675520396894,
      "loss": 1.3903,
      "step": 3995
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.000149369866435155,
      "loss": 1.3637,
      "step": 4000
    },
    {
      "epoch": 1.02,
      "eval_loss": 1.5145312547683716,
      "eval_runtime": 2.0288,
      "eval_samples_per_second": 49.29,
      "eval_steps_per_second": 1.972,
      "step": 4000
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0001492528884042129,
      "loss": 1.4221,
      "step": 4005
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00014913582132264222,
      "loss": 1.4648,
      "step": 4010
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00014901866540210356,
      "loss": 1.4487,
      "step": 4015
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00014890142085441824,
      "loss": 1.4623,
      "step": 4020
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0001487840878915677,
      "loss": 1.3967,
      "step": 4025
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00014866666672569328,
      "loss": 1.3644,
      "step": 4030
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00014854915756909586,
      "loss": 1.4131,
      "step": 4035
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00014843156063423527,
      "loss": 1.4199,
      "step": 4040
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00014831387613373017,
      "loss": 1.4172,
      "step": 4045
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00014819610428035753,
      "loss": 1.4314,
      "step": 4050
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0001480782452870522,
      "loss": 1.4041,
      "step": 4055
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00014796029936690655,
      "loss": 1.4303,
      "step": 4060
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00014784226673317022,
      "loss": 1.4184,
      "step": 4065
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00014772414759924956,
      "loss": 1.4041,
      "step": 4070
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00014760594217870737,
      "loss": 1.401,
      "step": 4075
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00014751131585951993,
      "loss": 1.4062,
      "step": 4080
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.00014739295566173482,
      "loss": 1.4252,
      "step": 4085
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.00014727450977613274,
      "loss": 1.4373,
      "step": 4090
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.00014715597841686708,
      "loss": 1.4052,
      "step": 4095
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.00014703736179824597,
      "loss": 1.4319,
      "step": 4100
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.00014691866013473166,
      "loss": 1.4012,
      "step": 4105
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0001467998736409401,
      "loss": 1.3806,
      "step": 4110
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0001466810025316408,
      "loss": 1.4366,
      "step": 4115
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.00014656204702175597,
      "loss": 1.4409,
      "step": 4120
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0001464430073263606,
      "loss": 1.4041,
      "step": 4125
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00014632388366068191,
      "loss": 1.4567,
      "step": 4130
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00014620467624009884,
      "loss": 1.4992,
      "step": 4135
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0001460853852801418,
      "loss": 1.4411,
      "step": 4140
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00014596601099649232,
      "loss": 1.4584,
      "step": 4145
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00014587045172154828,
      "loss": 1.4097,
      "step": 4150
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00014575092799925022,
      "loss": 1.4295,
      "step": 4155
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.000145631321557968,
      "loss": 1.4266,
      "step": 4160
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0001455116326139535,
      "loss": 1.3929,
      "step": 4165
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00014539186138360768,
      "loss": 1.4545,
      "step": 4170
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0001452720080834804,
      "loss": 1.3649,
      "step": 4175
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00014515207293026985,
      "loss": 1.4152,
      "step": 4180
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0001450320561408222,
      "loss": 1.4012,
      "step": 4185
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0001449119579321312,
      "loss": 1.3889,
      "step": 4190
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00014479177852133788,
      "loss": 1.4516,
      "step": 4195
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00014467151812573006,
      "loss": 1.3974,
      "step": 4200
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0001445511769627419,
      "loss": 1.4095,
      "step": 4205
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00014443075524995372,
      "loss": 1.4249,
      "step": 4210
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00014431025320509142,
      "loss": 1.418,
      "step": 4215
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0001441896710460261,
      "loss": 1.398,
      "step": 4220
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0001440690089907738,
      "loss": 1.3797,
      "step": 4225
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00014394826725749486,
      "loss": 1.4703,
      "step": 4230
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00014382744606449387,
      "loss": 1.4441,
      "step": 4235
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.000143706545630219,
      "loss": 1.4367,
      "step": 4240
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.00014358556617326162,
      "loss": 1.415,
      "step": 4245
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0001434645079123561,
      "loss": 1.4756,
      "step": 4250
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.00014334337106637922,
      "loss": 1.4432,
      "step": 4255
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.00014322215585434983,
      "loss": 1.4235,
      "step": 4260
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.00014310086249542852,
      "loss": 1.4393,
      "step": 4265
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.00014297949120891718,
      "loss": 1.4237,
      "step": 4270
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0001428580422142585,
      "loss": 1.3877,
      "step": 4275
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00014273651573103573,
      "loss": 1.4009,
      "step": 4280
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00014261491197897227,
      "loss": 1.4401,
      "step": 4285
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00014249323117793115,
      "loss": 1.3907,
      "step": 4290
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00014237147354791474,
      "loss": 1.3885,
      "step": 4295
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00014224963930906426,
      "loss": 1.3705,
      "step": 4300
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00014212772868165958,
      "loss": 1.4345,
      "step": 4305
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00014200574188611854,
      "loss": 1.5228,
      "step": 4310
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00014188367914299676,
      "loss": 1.4026,
      "step": 4315
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00014176154067298716,
      "loss": 1.4184,
      "step": 4320
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0001416393266969196,
      "loss": 1.4628,
      "step": 4325
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00014151703743576043,
      "loss": 1.4257,
      "step": 4330
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00014139467311061215,
      "loss": 1.4134,
      "step": 4335
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0001412722339427129,
      "loss": 1.4334,
      "step": 4340
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0001411497201534362,
      "loss": 1.4083,
      "step": 4345
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00014102713196429055,
      "loss": 1.4254,
      "step": 4350
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0001409044695969188,
      "loss": 1.4232,
      "step": 4355
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.00014078173327309806,
      "loss": 1.4508,
      "step": 4360
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0001406589232147391,
      "loss": 1.4353,
      "step": 4365
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0001405606222283936,
      "loss": 1.3598,
      "step": 4370
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0001404376800075063,
      "loss": 1.4732,
      "step": 4375
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.00014031466467413891,
      "loss": 1.5183,
      "step": 4380
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.00014019157645070665,
      "loss": 1.3976,
      "step": 4385
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0001400684155597566,
      "loss": 1.4479,
      "step": 4390
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00013994518222396713,
      "loss": 1.4417,
      "step": 4395
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00013982187666614762,
      "loss": 1.4082,
      "step": 4400
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00013969849910923802,
      "loss": 1.4277,
      "step": 4405
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00013957504977630853,
      "loss": 1.4101,
      "step": 4410
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00013945152889055902,
      "loss": 1.4257,
      "step": 4415
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0001393279366753188,
      "loss": 1.4512,
      "step": 4420
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0001392042733540462,
      "loss": 1.4347,
      "step": 4425
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00013908053915032785,
      "loss": 1.4204,
      "step": 4430
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0001389567342878789,
      "loss": 1.3492,
      "step": 4435
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.000138832858990542,
      "loss": 1.4595,
      "step": 4440
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.00013870891348228726,
      "loss": 1.4881,
      "step": 4445
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0001385848979872116,
      "loss": 1.4345,
      "step": 4450
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.00013846081272953867,
      "loss": 1.4488,
      "step": 4455
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.00013833665793361818,
      "loss": 1.3982,
      "step": 4460
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.00013821243382392548,
      "loss": 1.4072,
      "step": 4465
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.00013811300478118284,
      "loss": 1.4407,
      "step": 4470
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00013798865647277801,
      "loss": 1.449,
      "step": 4475
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0001378891283623639,
      "loss": 1.3961,
      "step": 4480
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0001377646565836807,
      "loss": 1.4503,
      "step": 4485
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00013764011652542003,
      "loss": 1.4568,
      "step": 4490
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00013751550841275393,
      "loss": 1.4734,
      "step": 4495
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00013739083247097747,
      "loss": 1.4814,
      "step": 4500
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00013726608892550822,
      "loss": 1.4997,
      "step": 4505
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00013714127800188613,
      "loss": 1.4476,
      "step": 4510
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00013704138090235902,
      "loss": 1.4278,
      "step": 4515
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0001369164492668109,
      "loss": 1.409,
      "step": 4520
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00013679145088526842,
      "loss": 1.3598,
      "step": 4525
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00013666638598373225,
      "loss": 1.4,
      "step": 4530
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00013654125478832326,
      "loss": 1.4285,
      "step": 4535
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00013641605752528224,
      "loss": 1.4639,
      "step": 4540
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0001362907944209694,
      "loss": 1.4717,
      "step": 4545
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.000136165465701864,
      "loss": 1.407,
      "step": 4550
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.000136040071594564,
      "loss": 1.4288,
      "step": 4555
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00013591461232578545,
      "loss": 1.4438,
      "step": 4560
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0001357890881223624,
      "loss": 1.4144,
      "step": 4565
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00013566349921124608,
      "loss": 1.4002,
      "step": 4570
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00013553784581950485,
      "loss": 1.446,
      "step": 4575
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0001354121281743237,
      "loss": 1.4575,
      "step": 4580
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00013528634650300365,
      "loss": 1.401,
      "step": 4585
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00013516050103296157,
      "loss": 1.4169,
      "step": 4590
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00013503459199172967,
      "loss": 1.4195,
      "step": 4595
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0001349086196069551,
      "loss": 1.4326,
      "step": 4600
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00013478258410639952,
      "loss": 1.4582,
      "step": 4605
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0001346564857179387,
      "loss": 1.4867,
      "step": 4610
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00013453032466956218,
      "loss": 1.4168,
      "step": 4615
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00013440410118937266,
      "loss": 1.3966,
      "step": 4620
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00013427781550558586,
      "loss": 1.3888,
      "step": 4625
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00013415146784652985,
      "loss": 1.5074,
      "step": 4630
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00013402505844064487,
      "loss": 1.4581,
      "step": 4635
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00013389858751648265,
      "loss": 1.4293,
      "step": 4640
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0001337720553027063,
      "loss": 1.4107,
      "step": 4645
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00013364546202808964,
      "loss": 1.3778,
      "step": 4650
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.000133518807921517,
      "loss": 1.3992,
      "step": 4655
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00013339209321198254,
      "loss": 1.4125,
      "step": 4660
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00013326531812859002,
      "loss": 1.4158,
      "step": 4665
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.00013313848290055256,
      "loss": 1.4829,
      "step": 4670
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0001330115877571918,
      "loss": 1.4541,
      "step": 4675
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.00013288463292793776,
      "loss": 1.4877,
      "step": 4680
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0001327576186423284,
      "loss": 1.4225,
      "step": 4685
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0001326305451300092,
      "loss": 1.4739,
      "step": 4690
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.00013250341262073274,
      "loss": 1.4001,
      "step": 4695
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.00013237622134435815,
      "loss": 1.4394,
      "step": 4700
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0001322489715308509,
      "loss": 1.4119,
      "step": 4705
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0001321216634102823,
      "loss": 1.3745,
      "step": 4710
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0001319942972128291,
      "loss": 1.4752,
      "step": 4715
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00013186687316877298,
      "loss": 1.4181,
      "step": 4720
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00013173939150850028,
      "loss": 1.4892,
      "step": 4725
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00013161185246250142,
      "loss": 1.4416,
      "step": 4730
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0001314842562613707,
      "loss": 1.4177,
      "step": 4735
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0001313566031358057,
      "loss": 1.4313,
      "step": 4740
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00013122889331660688,
      "loss": 1.3827,
      "step": 4745
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00013110112703467724,
      "loss": 1.4574,
      "step": 4750
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0001309733045210219,
      "loss": 1.4436,
      "step": 4755
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00013084542600674755,
      "loss": 1.4202,
      "step": 4760
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00013071749172306228,
      "loss": 1.424,
      "step": 4765
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00013058950190127487,
      "loss": 1.4608,
      "step": 4770
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0001304614567727946,
      "loss": 1.3907,
      "step": 4775
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00013035898100476238,
      "loss": 1.4107,
      "step": 4780
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00013023083690770517,
      "loss": 1.4548,
      "step": 4785
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.00013010263815243164,
      "loss": 1.4549,
      "step": 4790
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.00013000003995005462,
      "loss": 1.4208,
      "step": 4795
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0001298717433941611,
      "loss": 1.4437,
      "step": 4800
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0001297433928293026,
      "loss": 1.4486,
      "step": 4805
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.00012961498848754058,
      "loss": 1.407,
      "step": 4810
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.00012948653060103376,
      "loss": 1.3718,
      "step": 4815
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0001293580194020377,
      "loss": 1.4805,
      "step": 4820
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00012922945512290428,
      "loss": 1.45,
      "step": 4825
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00012910083799608137,
      "loss": 1.4157,
      "step": 4830
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00012897216825411245,
      "loss": 1.3946,
      "step": 4835
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00012884344612963607,
      "loss": 1.4229,
      "step": 4840
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00012871467185538552,
      "loss": 1.4577,
      "step": 4845
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00012858584566418838,
      "loss": 1.4174,
      "step": 4850
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00012845696778896598,
      "loss": 1.4712,
      "step": 4855
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0001283280384627333,
      "loss": 1.4475,
      "step": 4860
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00012819905791859829,
      "loss": 1.3623,
      "step": 4865
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00012807002638976132,
      "loss": 1.418,
      "step": 4870
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00012794094410951515,
      "loss": 1.3749,
      "step": 4875
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0001278118113112442,
      "loss": 1.4437,
      "step": 4880
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0001276826282284243,
      "loss": 1.4181,
      "step": 4885
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00012755339509462214,
      "loss": 1.384,
      "step": 4890
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00012742411214349488,
      "loss": 1.39,
      "step": 4895
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0001273206500711942,
      "loss": 1.4391,
      "step": 4900
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00012719127803798518,
      "loss": 1.4204,
      "step": 4905
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00012706185684216906,
      "loss": 1.3879,
      "step": 4910
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00012693238671774307,
      "loss": 1.4231,
      "step": 4915
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0001268028678987929,
      "loss": 1.3655,
      "step": 4920
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00012667330061949224,
      "loss": 1.4627,
      "step": 4925
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00012654368511410245,
      "loss": 1.4352,
      "step": 4930
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00012641402161697206,
      "loss": 1.4179,
      "step": 4935
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0001262843103625364,
      "loss": 1.4422,
      "step": 4940
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00012618050713132368,
      "loss": 1.3732,
      "step": 4945
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00012605071050479077,
      "loss": 1.4328,
      "step": 4950
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00012592086677782946,
      "loss": 1.357,
      "step": 4955
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00012579097618520087,
      "loss": 1.4299,
      "step": 4960
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0001256610389617509,
      "loss": 1.4426,
      "step": 4965
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00012555705576666986,
      "loss": 1.4366,
      "step": 4970
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00012542703519982198,
      "loss": 1.4408,
      "step": 4975
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0001252969686601684,
      "loss": 1.4207,
      "step": 4980
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00012516685638287319,
      "loss": 1.3681,
      "step": 4985
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.000125036698603183,
      "loss": 1.3886,
      "step": 4990
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00012490649555642693,
      "loss": 1.3916,
      "step": 4995
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00012477624747801567,
      "loss": 1.4379,
      "step": 5000
    },
    {
      "epoch": 1.28,
      "eval_loss": 1.4501123428344727,
      "eval_runtime": 2.0229,
      "eval_samples_per_second": 49.433,
      "eval_steps_per_second": 1.977,
      "step": 5000
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00012464595460344154,
      "loss": 1.3558,
      "step": 5005
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00012451561716827778,
      "loss": 1.4617,
      "step": 5010
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00012438523540817816,
      "loss": 1.4486,
      "step": 5015
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00012425480955887663,
      "loss": 1.4118,
      "step": 5020
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00012412433985618685,
      "loss": 1.4532,
      "step": 5025
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0001239938265360018,
      "loss": 1.406,
      "step": 5030
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00012386326983429327,
      "loss": 1.4404,
      "step": 5035
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0001237326699871115,
      "loss": 1.4229,
      "step": 5040
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00012360202723058473,
      "loss": 1.3981,
      "step": 5045
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0001234713418009188,
      "loss": 1.4176,
      "step": 5050
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00012334061393439675,
      "loss": 1.4095,
      "step": 5055
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00012320984386737823,
      "loss": 1.4348,
      "step": 5060
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00012307903183629928,
      "loss": 1.4238,
      "step": 5065
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00012294817807767175,
      "loss": 1.3832,
      "step": 5070
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00012281728282808306,
      "loss": 1.4585,
      "step": 5075
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0001226863463241955,
      "loss": 1.4286,
      "step": 5080
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00012255536880274596,
      "loss": 1.4371,
      "step": 5085
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00012242435050054563,
      "loss": 1.3724,
      "step": 5090
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0001222932916544793,
      "loss": 1.389,
      "step": 5095
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00012216219250150512,
      "loss": 1.4051,
      "step": 5100
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00012203105327865407,
      "loss": 1.4805,
      "step": 5105
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00012189987422302964,
      "loss": 1.4422,
      "step": 5110
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0001217686555718073,
      "loss": 1.4036,
      "step": 5115
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00012163739756223411,
      "loss": 1.344,
      "step": 5120
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00012150610043162832,
      "loss": 1.4108,
      "step": 5125
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00012137476441737888,
      "loss": 1.4028,
      "step": 5130
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00012124338975694503,
      "loss": 1.4541,
      "step": 5135
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00012111197668785592,
      "loss": 1.4288,
      "step": 5140
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00012098052544771019,
      "loss": 1.4387,
      "step": 5145
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00012084903627417535,
      "loss": 1.4569,
      "step": 5150
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00012071750940498761,
      "loss": 1.4091,
      "step": 5155
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00012058594507795131,
      "loss": 1.4269,
      "step": 5160
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00012045434353093851,
      "loss": 1.3892,
      "step": 5165
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00012032270500188858,
      "loss": 1.427,
      "step": 5170
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00012019102972880774,
      "loss": 1.4119,
      "step": 5175
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00012005931794976859,
      "loss": 1.4303,
      "step": 5180
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00011992756990290984,
      "loss": 1.4263,
      "step": 5185
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00011979578582643569,
      "loss": 1.4554,
      "step": 5190
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00011966396595861555,
      "loss": 1.4347,
      "step": 5195
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00011953211053778351,
      "loss": 1.4639,
      "step": 5200
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00011945298031900956,
      "loss": 1.4315,
      "step": 5205
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0001193210685092536,
      "loss": 1.3672,
      "step": 5210
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0001191891217664541,
      "loss": 1.383,
      "step": 5215
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00011905714032917448,
      "loss": 1.4325,
      "step": 5220
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00011892512443604102,
      "loss": 1.411,
      "step": 5225
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00011879307432574223,
      "loss": 1.3938,
      "step": 5230
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00011866099023702841,
      "loss": 1.4079,
      "step": 5235
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00011852887240871145,
      "loss": 1.4273,
      "step": 5240
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00011839672107966406,
      "loss": 1.3616,
      "step": 5245
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00011826453648881972,
      "loss": 1.3725,
      "step": 5250
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00011813231887517189,
      "loss": 1.3689,
      "step": 5255
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00011800006847777377,
      "loss": 1.3701,
      "step": 5260
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00011786778553573787,
      "loss": 1.3809,
      "step": 5265
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00011773547028823557,
      "loss": 1.4298,
      "step": 5270
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00011760312297449656,
      "loss": 1.3978,
      "step": 5275
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0001174707438338086,
      "loss": 1.3769,
      "step": 5280
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00011733833310551692,
      "loss": 1.3599,
      "step": 5285
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0001172058910290239,
      "loss": 1.4283,
      "step": 5290
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00011707341784378864,
      "loss": 1.4419,
      "step": 5295
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00011694091378932641,
      "loss": 1.3855,
      "step": 5300
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00011680837910520834,
      "loss": 1.458,
      "step": 5305
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00011667581403106089,
      "loss": 1.3965,
      "step": 5310
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00011654321880656552,
      "loss": 1.4395,
      "step": 5315
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00011641059367145819,
      "loss": 1.4491,
      "step": 5320
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00011630447218886882,
      "loss": 1.4088,
      "step": 5325
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00011617179381896688,
      "loss": 1.43,
      "step": 5330
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.00011603908620999997,
      "loss": 1.4521,
      "step": 5335
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.00011590634960190721,
      "loss": 1.3744,
      "step": 5340
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.00011577358423468021,
      "loss": 1.4356,
      "step": 5345
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0001156407903483625,
      "loss": 1.4306,
      "step": 5350
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.00011550796818304925,
      "loss": 1.4369,
      "step": 5355
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.00011540169025129845,
      "loss": 1.3815,
      "step": 5360
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.00011526881778899591,
      "loss": 1.431,
      "step": 5365
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.00011513591772023478,
      "loss": 1.4024,
      "step": 5370
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00011500299028530228,
      "loss": 1.4084,
      "step": 5375
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00011487003572453496,
      "loss": 1.4309,
      "step": 5380
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00011473705427831842,
      "loss": 1.4089,
      "step": 5385
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00011460404618708698,
      "loss": 1.4052,
      "step": 5390
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00011447101169132302,
      "loss": 1.425,
      "step": 5395
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00011433795103155676,
      "loss": 1.3861,
      "step": 5400
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00011420486444836562,
      "loss": 1.4156,
      "step": 5405
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00011407175218237394,
      "loss": 1.4007,
      "step": 5410
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00011393861447425252,
      "loss": 1.3826,
      "step": 5415
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00011380545156471815,
      "loss": 1.4594,
      "step": 5420
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00011367226369453316,
      "loss": 1.3712,
      "step": 5425
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00011353905110450503,
      "loss": 1.4281,
      "step": 5430
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00011340581403548589,
      "loss": 1.4149,
      "step": 5435
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00011327255272837221,
      "loss": 1.3817,
      "step": 5440
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00011313926742410421,
      "loss": 1.4171,
      "step": 5445
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.00011300595836366555,
      "loss": 1.4425,
      "step": 5450
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0001128726257880828,
      "loss": 1.3551,
      "step": 5455
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.00011273926993842505,
      "loss": 1.3763,
      "step": 5460
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.00011260589105580352,
      "loss": 1.3562,
      "step": 5465
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.00011247248938137103,
      "loss": 1.3517,
      "step": 5470
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.00011233906515632158,
      "loss": 1.4338,
      "step": 5475
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.00011223230970194626,
      "loss": 1.4113,
      "step": 5480
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0001120988454937252,
      "loss": 1.4015,
      "step": 5485
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00011196535941044599,
      "loss": 1.4032,
      "step": 5490
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00011183185169345534,
      "loss": 1.4157,
      "step": 5495
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00011169832258413897,
      "loss": 1.4636,
      "step": 5500
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00011156477232392133,
      "loss": 1.3721,
      "step": 5505
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00011145791704936,
      "loss": 1.4165,
      "step": 5510
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00011132432932603147,
      "loss": 1.4159,
      "step": 5515
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00011119072112799218,
      "loss": 1.3732,
      "step": 5520
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0001110570926968096,
      "loss": 1.3909,
      "step": 5525
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00011092344427408767,
      "loss": 1.4006,
      "step": 5530
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00011078977610146661,
      "loss": 1.3661,
      "step": 5535
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00011065608842062226,
      "loss": 1.4046,
      "step": 5540
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00011052238147326575,
      "loss": 1.4093,
      "step": 5545
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00011038865550114314,
      "loss": 1.4184,
      "step": 5550
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00011025491074603473,
      "loss": 1.4129,
      "step": 5555
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00011012114744975488,
      "loss": 1.442,
      "step": 5560
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00010998736585415143,
      "loss": 1.4422,
      "step": 5565
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00010985356620110534,
      "loss": 1.3884,
      "step": 5570
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00010971974873253024,
      "loss": 1.4395,
      "step": 5575
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0001095859136903719,
      "loss": 1.4436,
      "step": 5580
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00010945206131660786,
      "loss": 1.4015,
      "step": 5585
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0001093181918532471,
      "loss": 1.4198,
      "step": 5590
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0001091843055423294,
      "loss": 1.4142,
      "step": 5595
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00010905040262592501,
      "loss": 1.5011,
      "step": 5600
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00010891648334613427,
      "loss": 1.3874,
      "step": 5605
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.00010878254794508704,
      "loss": 1.3769,
      "step": 5610
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0001086753881796745,
      "loss": 1.458,
      "step": 5615
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.00010854142437062567,
      "loss": 1.4166,
      "step": 5620
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.00010840744511843768,
      "loss": 1.3609,
      "step": 5625
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.00010827345066534887,
      "loss": 1.4003,
      "step": 5630
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0001081662443210309,
      "loss": 1.4183,
      "step": 5635
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.00010803222311684858,
      "loss": 1.4088,
      "step": 5640
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.00010792499568567884,
      "loss": 1.4285,
      "step": 5645
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.00010779094851550066,
      "loss": 1.4227,
      "step": 5650
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.00010765688725906506,
      "loss": 1.3918,
      "step": 5655
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.00010752281215875871,
      "loss": 1.4503,
      "step": 5660
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.00010738872345699313,
      "loss": 1.4094,
      "step": 5665
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.00010725462139620459,
      "loss": 1.3559,
      "step": 5670
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.00010712050621885343,
      "loss": 1.3442,
      "step": 5675
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0001069863781674237,
      "loss": 1.3906,
      "step": 5680
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00010685223748442275,
      "loss": 1.3976,
      "step": 5685
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00010671808441238078,
      "loss": 1.3859,
      "step": 5690
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00010658391919385035,
      "loss": 1.4548,
      "step": 5695
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00010644974207140608,
      "loss": 1.4269,
      "step": 5700
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00010631555328764393,
      "loss": 1.4029,
      "step": 5705
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00010618135308518111,
      "loss": 1.4477,
      "step": 5710
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00010604714170665544,
      "loss": 1.3715,
      "step": 5715
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00010591291939472487,
      "loss": 1.4253,
      "step": 5720
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.00010577868639206722,
      "loss": 1.3838,
      "step": 5725
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.00010564444294137956,
      "loss": 1.3944,
      "step": 5730
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0001055101892853779,
      "loss": 1.3913,
      "step": 5735
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.00010537592566679669,
      "loss": 1.3901,
      "step": 5740
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.00010524165232838836,
      "loss": 1.4328,
      "step": 5745
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.00010510736951292295,
      "loss": 1.3933,
      "step": 5750
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.00010497307746318762,
      "loss": 1.4321,
      "step": 5755
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0001048656373378887,
      "loss": 1.4344,
      "step": 5760
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0001047313292783442,
      "loss": 1.4141,
      "step": 5765
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.00010459701266442126,
      "loss": 1.3427,
      "step": 5770
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.00010446268773896817,
      "loss": 1.3734,
      "step": 5775
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.00010432835474484825,
      "loss": 1.4039,
      "step": 5780
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.00010419401392493935,
      "loss": 1.3891,
      "step": 5785
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.00010405966552213351,
      "loss": 1.4056,
      "step": 5790
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.00010392530977933648,
      "loss": 1.3927,
      "step": 5795
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.00010379094693946732,
      "loss": 1.3752,
      "step": 5800
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00010368345172092979,
      "loss": 1.3876,
      "step": 5805
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00010357595224012906,
      "loss": 1.3996,
      "step": 5810
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0001034415720848241,
      "loss": 1.3784,
      "step": 5815
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00010330718570705853,
      "loss": 1.374,
      "step": 5820
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00010317279334980678,
      "loss": 1.3702,
      "step": 5825
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00010303839525605405,
      "loss": 1.3943,
      "step": 5830
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00010290399166879594,
      "loss": 1.4236,
      "step": 5835
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00010276958283103797,
      "loss": 1.3874,
      "step": 5840
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00010263516898579517,
      "loss": 1.3695,
      "step": 5845
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00010250075037609161,
      "loss": 1.3526,
      "step": 5850
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00010236632724495993,
      "loss": 1.3377,
      "step": 5855
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00010223189983544104,
      "loss": 1.3969,
      "step": 5860
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00010209746839058354,
      "loss": 1.3599,
      "step": 5865
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00010196303315344326,
      "loss": 1.4172,
      "step": 5870
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00010182859436708303,
      "loss": 1.4567,
      "step": 5875
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00010169415227457196,
      "loss": 1.414,
      "step": 5880
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00010155970711898519,
      "loss": 1.3855,
      "step": 5885
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00010142525914340344,
      "loss": 1.3983,
      "step": 5890
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00010129080859091242,
      "loss": 1.3987,
      "step": 5895
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00010115635570460262,
      "loss": 1.4195,
      "step": 5900
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00010102190072756872,
      "loss": 1.4462,
      "step": 5905
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0001008874439029091,
      "loss": 1.4056,
      "step": 5910
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0001007529854737256,
      "loss": 1.4045,
      "step": 5915
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.00010061852568312282,
      "loss": 1.378,
      "step": 5920
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.00010048406477420795,
      "loss": 1.4047,
      "step": 5925
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.00010034960299009018,
      "loss": 1.4246,
      "step": 5930
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.00010021514057388019,
      "loss": 1.4401,
      "step": 5935
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0001000806777686899,
      "loss": 1.4515,
      "step": 5940
    },
    {
      "epoch": 1.52,
      "learning_rate": 9.994621481763191e-05,
      "loss": 1.3925,
      "step": 5945
    },
    {
      "epoch": 1.52,
      "learning_rate": 9.981175196381905e-05,
      "loss": 1.4343,
      "step": 5950
    },
    {
      "epoch": 1.52,
      "learning_rate": 9.967728945036407e-05,
      "loss": 1.4574,
      "step": 5955
    },
    {
      "epoch": 1.53,
      "learning_rate": 9.954282752037893e-05,
      "loss": 1.3839,
      "step": 5960
    },
    {
      "epoch": 1.53,
      "learning_rate": 9.940836641697465e-05,
      "loss": 1.4161,
      "step": 5965
    },
    {
      "epoch": 1.53,
      "learning_rate": 9.927390638326081e-05,
      "loss": 1.3791,
      "step": 5970
    },
    {
      "epoch": 1.53,
      "learning_rate": 9.913944766234502e-05,
      "loss": 1.4105,
      "step": 5975
    },
    {
      "epoch": 1.53,
      "learning_rate": 9.900499049733238e-05,
      "loss": 1.419,
      "step": 5980
    },
    {
      "epoch": 1.53,
      "learning_rate": 9.887053513132533e-05,
      "loss": 1.4007,
      "step": 5985
    },
    {
      "epoch": 1.53,
      "learning_rate": 9.873608180742302e-05,
      "loss": 1.4211,
      "step": 5990
    },
    {
      "epoch": 1.53,
      "learning_rate": 9.860163076872089e-05,
      "loss": 1.3887,
      "step": 5995
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.846718225831024e-05,
      "loss": 1.3633,
      "step": 6000
    },
    {
      "epoch": 1.54,
      "eval_loss": 1.4240084886550903,
      "eval_runtime": 2.0288,
      "eval_samples_per_second": 49.289,
      "eval_steps_per_second": 1.972,
      "step": 6000
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.833273651927786e-05,
      "loss": 1.4191,
      "step": 6005
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.819829379470545e-05,
      "loss": 1.3797,
      "step": 6010
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.80638543276693e-05,
      "loss": 1.4165,
      "step": 6015
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.792941836123985e-05,
      "loss": 1.4234,
      "step": 6020
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.779498613848109e-05,
      "loss": 1.4159,
      "step": 6025
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.766055790245034e-05,
      "loss": 1.3976,
      "step": 6030
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.752613389619772e-05,
      "loss": 1.3828,
      "step": 6035
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.739171436276564e-05,
      "loss": 1.376,
      "step": 6040
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.725729954518847e-05,
      "loss": 1.4632,
      "step": 6045
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.712288968649202e-05,
      "loss": 1.3897,
      "step": 6050
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.69884850296932e-05,
      "loss": 1.3601,
      "step": 6055
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.685408581779942e-05,
      "loss": 1.3986,
      "step": 6060
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.671969229380829e-05,
      "loss": 1.3543,
      "step": 6065
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.658530470070722e-05,
      "loss": 1.4197,
      "step": 6070
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.64509232814727e-05,
      "loss": 1.4355,
      "step": 6075
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.631654827907021e-05,
      "loss": 1.3688,
      "step": 6080
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.618217993645367e-05,
      "loss": 1.3824,
      "step": 6085
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.604781849656477e-05,
      "loss": 1.388,
      "step": 6090
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.591346420233286e-05,
      "loss": 1.4368,
      "step": 6095
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.57791172966744e-05,
      "loss": 1.4085,
      "step": 6100
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.564477802249236e-05,
      "loss": 1.3496,
      "step": 6105
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.551044662267602e-05,
      "loss": 1.3607,
      "step": 6110
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.53761233401004e-05,
      "loss": 1.4359,
      "step": 6115
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.526867072165581e-05,
      "loss": 1.3941,
      "step": 6120
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.513436266211135e-05,
      "loss": 1.3779,
      "step": 6125
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.500006339977808e-05,
      "loss": 1.3884,
      "step": 6130
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.489263048707706e-05,
      "loss": 1.345,
      "step": 6135
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.475834767161165e-05,
      "loss": 1.3983,
      "step": 6140
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.462407433320335e-05,
      "loss": 1.3893,
      "step": 6145
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.448981071462211e-05,
      "loss": 1.4511,
      "step": 6150
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.438240698116291e-05,
      "loss": 1.3767,
      "step": 6155
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.424816146999457e-05,
      "loss": 1.3907,
      "step": 6160
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.411392635831452e-05,
      "loss": 1.4223,
      "step": 6165
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.397970188882373e-05,
      "loss": 1.3985,
      "step": 6170
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.384548830420376e-05,
      "loss": 1.4028,
      "step": 6175
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.371128584711669e-05,
      "loss": 1.4084,
      "step": 6180
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.357709476020442e-05,
      "loss": 1.4113,
      "step": 6185
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.344291528608818e-05,
      "loss": 1.3479,
      "step": 6190
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.330874766736832e-05,
      "loss": 1.3876,
      "step": 6195
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.317459214662374e-05,
      "loss": 1.3931,
      "step": 6200
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.304044896641147e-05,
      "loss": 1.3901,
      "step": 6205
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.290631836926615e-05,
      "loss": 1.4116,
      "step": 6210
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.277220059769979e-05,
      "loss": 1.3667,
      "step": 6215
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.263809589420114e-05,
      "loss": 1.3717,
      "step": 6220
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.25040045012353e-05,
      "loss": 1.386,
      "step": 6225
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.236992666124335e-05,
      "loss": 1.3505,
      "step": 6230
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.223586261664192e-05,
      "loss": 1.3537,
      "step": 6235
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.210181260982252e-05,
      "loss": 1.3243,
      "step": 6240
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.196777688315145e-05,
      "loss": 1.3467,
      "step": 6245
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.183375567896912e-05,
      "loss": 1.3835,
      "step": 6250
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.16997492395897e-05,
      "loss": 1.3783,
      "step": 6255
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.156575780730064e-05,
      "loss": 1.3511,
      "step": 6260
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.143178162436232e-05,
      "loss": 1.3788,
      "step": 6265
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.13246118203255e-05,
      "loss": 1.425,
      "step": 6270
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.119066369662623e-05,
      "loss": 1.3933,
      "step": 6275
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.10567315004585e-05,
      "loss": 1.3429,
      "step": 6280
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.092281547397559e-05,
      "loss": 1.3718,
      "step": 6285
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.07889158593015e-05,
      "loss": 1.4122,
      "step": 6290
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.06550328985305e-05,
      "loss": 1.3429,
      "step": 6295
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.052116683372688e-05,
      "loss": 1.3735,
      "step": 6300
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.038731790692428e-05,
      "loss": 1.3651,
      "step": 6305
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.025348636012536e-05,
      "loss": 1.3558,
      "step": 6310
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.01196724353014e-05,
      "loss": 1.3307,
      "step": 6315
    },
    {
      "epoch": 1.62,
      "learning_rate": 8.998587637439185e-05,
      "loss": 1.3343,
      "step": 6320
    },
    {
      "epoch": 1.62,
      "learning_rate": 8.985209841930369e-05,
      "loss": 1.3435,
      "step": 6325
    },
    {
      "epoch": 1.62,
      "learning_rate": 8.971833881191134e-05,
      "loss": 1.362,
      "step": 6330
    },
    {
      "epoch": 1.62,
      "learning_rate": 8.958459779405601e-05,
      "loss": 1.3608,
      "step": 6335
    },
    {
      "epoch": 1.62,
      "learning_rate": 8.94508756075453e-05,
      "loss": 1.3742,
      "step": 6340
    },
    {
      "epoch": 1.62,
      "learning_rate": 8.931717249415267e-05,
      "loss": 1.3826,
      "step": 6345
    },
    {
      "epoch": 1.63,
      "learning_rate": 8.918348869561721e-05,
      "loss": 1.3985,
      "step": 6350
    },
    {
      "epoch": 1.63,
      "learning_rate": 8.904982445364303e-05,
      "loss": 1.3799,
      "step": 6355
    },
    {
      "epoch": 1.63,
      "learning_rate": 8.891618000989891e-05,
      "loss": 1.3598,
      "step": 6360
    },
    {
      "epoch": 1.63,
      "learning_rate": 8.878255560601781e-05,
      "loss": 1.3633,
      "step": 6365
    },
    {
      "epoch": 1.63,
      "learning_rate": 8.864895148359647e-05,
      "loss": 1.3742,
      "step": 6370
    },
    {
      "epoch": 1.63,
      "learning_rate": 8.851536788419495e-05,
      "loss": 1.3781,
      "step": 6375
    },
    {
      "epoch": 1.63,
      "learning_rate": 8.840851594355273e-05,
      "loss": 1.3448,
      "step": 6380
    },
    {
      "epoch": 1.63,
      "learning_rate": 8.827496989419954e-05,
      "loss": 1.3512,
      "step": 6385
    },
    {
      "epoch": 1.64,
      "learning_rate": 8.814144504403558e-05,
      "loss": 1.3487,
      "step": 6390
    },
    {
      "epoch": 1.64,
      "learning_rate": 8.800794163447763e-05,
      "loss": 1.3791,
      "step": 6395
    },
    {
      "epoch": 1.64,
      "learning_rate": 8.787445990690355e-05,
      "loss": 1.3459,
      "step": 6400
    },
    {
      "epoch": 1.64,
      "learning_rate": 8.776769029805376e-05,
      "loss": 1.4147,
      "step": 6405
    },
    {
      "epoch": 1.64,
      "learning_rate": 8.763424820619789e-05,
      "loss": 1.4235,
      "step": 6410
    },
    {
      "epoch": 1.64,
      "learning_rate": 8.750082847197436e-05,
      "loss": 1.3645,
      "step": 6415
    },
    {
      "epoch": 1.64,
      "learning_rate": 8.736743133660983e-05,
      "loss": 1.3669,
      "step": 6420
    },
    {
      "epoch": 1.64,
      "learning_rate": 8.723405704129015e-05,
      "loss": 1.3283,
      "step": 6425
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.710070582715987e-05,
      "loss": 1.3286,
      "step": 6430
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.696737793532175e-05,
      "loss": 1.3463,
      "step": 6435
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.683407360683644e-05,
      "loss": 1.4076,
      "step": 6440
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.670079308272199e-05,
      "loss": 1.3792,
      "step": 6445
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.656753660395344e-05,
      "loss": 1.3589,
      "step": 6450
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.64343044114622e-05,
      "loss": 1.3516,
      "step": 6455
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.63010967461359e-05,
      "loss": 1.3236,
      "step": 6460
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.616791384881791e-05,
      "loss": 1.3945,
      "step": 6465
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.603475596030658e-05,
      "loss": 1.4158,
      "step": 6470
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.592824781762609e-05,
      "loss": 1.3717,
      "step": 6475
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.579513555163441e-05,
      "loss": 1.4276,
      "step": 6480
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.566204896844325e-05,
      "loss": 1.3594,
      "step": 6485
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.5528988308677e-05,
      "loss": 1.3455,
      "step": 6490
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.539595381291304e-05,
      "loss": 1.3707,
      "step": 6495
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.526294572168156e-05,
      "loss": 1.3732,
      "step": 6500
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.512996427546509e-05,
      "loss": 1.3353,
      "step": 6505
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.499700971469774e-05,
      "loss": 1.3812,
      "step": 6510
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.486408227976521e-05,
      "loss": 1.379,
      "step": 6515
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.473118221100414e-05,
      "loss": 1.3487,
      "step": 6520
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.459830974870156e-05,
      "loss": 1.3838,
      "step": 6525
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.446546513309474e-05,
      "loss": 1.3404,
      "step": 6530
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.433264860437056e-05,
      "loss": 1.3956,
      "step": 6535
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.4199860402665e-05,
      "loss": 1.3947,
      "step": 6540
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.409365039809281e-05,
      "loss": 1.4088,
      "step": 6545
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.396091379000005e-05,
      "loss": 1.3934,
      "step": 6550
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.382820618103313e-05,
      "loss": 1.3734,
      "step": 6555
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.369552781113122e-05,
      "loss": 1.3308,
      "step": 6560
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.356287892018054e-05,
      "loss": 1.328,
      "step": 6565
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.343025974801412e-05,
      "loss": 1.3877,
      "step": 6570
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.329767053441124e-05,
      "loss": 1.4135,
      "step": 6575
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.316511151909689e-05,
      "loss": 1.3353,
      "step": 6580
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.30325829417416e-05,
      "loss": 1.4085,
      "step": 6585
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.290008504196088e-05,
      "loss": 1.4019,
      "step": 6590
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.276761805931463e-05,
      "loss": 1.3734,
      "step": 6595
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.263518223330697e-05,
      "loss": 1.386,
      "step": 6600
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.25027778033857e-05,
      "loss": 1.3633,
      "step": 6605
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.237040500894171e-05,
      "loss": 1.359,
      "step": 6610
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.223806408930885e-05,
      "loss": 1.3589,
      "step": 6615
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.21057552837633e-05,
      "loss": 1.3916,
      "step": 6620
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.197347883152315e-05,
      "loss": 1.3352,
      "step": 6625
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.184123497174798e-05,
      "loss": 1.3992,
      "step": 6630
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.170902394353848e-05,
      "loss": 1.3857,
      "step": 6635
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.157684598593597e-05,
      "loss": 1.3681,
      "step": 6640
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.144470133792194e-05,
      "loss": 1.3834,
      "step": 6645
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.131259023841772e-05,
      "loss": 1.3725,
      "step": 6650
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.118051292628394e-05,
      "loss": 1.379,
      "step": 6655
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.104846964032015e-05,
      "loss": 1.3235,
      "step": 6660
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.091646061926437e-05,
      "loss": 1.3643,
      "step": 6665
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.07844861017927e-05,
      "loss": 1.3295,
      "step": 6670
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.065254632651881e-05,
      "loss": 1.3032,
      "step": 6675
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.05206415319936e-05,
      "loss": 1.3784,
      "step": 6680
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.038877195670475e-05,
      "loss": 1.413,
      "step": 6685
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.025693783907613e-05,
      "loss": 1.3758,
      "step": 6690
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.012513941746761e-05,
      "loss": 1.3375,
      "step": 6695
    },
    {
      "epoch": 1.72,
      "learning_rate": 7.999337693017452e-05,
      "loss": 1.3502,
      "step": 6700
    },
    {
      "epoch": 1.72,
      "learning_rate": 7.986165061542727e-05,
      "loss": 1.4262,
      "step": 6705
    },
    {
      "epoch": 1.72,
      "learning_rate": 7.972996071139064e-05,
      "loss": 1.3947,
      "step": 6710
    },
    {
      "epoch": 1.72,
      "learning_rate": 7.959830745616386e-05,
      "loss": 1.4095,
      "step": 6715
    },
    {
      "epoch": 1.72,
      "learning_rate": 7.946669108777975e-05,
      "loss": 1.3522,
      "step": 6720
    },
    {
      "epoch": 1.72,
      "learning_rate": 7.933511184420443e-05,
      "loss": 1.3396,
      "step": 6725
    },
    {
      "epoch": 1.72,
      "learning_rate": 7.920356996333694e-05,
      "loss": 1.3092,
      "step": 6730
    },
    {
      "epoch": 1.72,
      "learning_rate": 7.907206568300879e-05,
      "loss": 1.3781,
      "step": 6735
    },
    {
      "epoch": 1.73,
      "learning_rate": 7.894059924098343e-05,
      "loss": 1.3828,
      "step": 6740
    },
    {
      "epoch": 1.73,
      "learning_rate": 7.880917087495595e-05,
      "loss": 1.3521,
      "step": 6745
    },
    {
      "epoch": 1.73,
      "learning_rate": 7.867778082255264e-05,
      "loss": 1.3787,
      "step": 6750
    },
    {
      "epoch": 1.73,
      "learning_rate": 7.85464293213304e-05,
      "loss": 1.4029,
      "step": 6755
    },
    {
      "epoch": 1.73,
      "learning_rate": 7.841511660877651e-05,
      "loss": 1.3751,
      "step": 6760
    },
    {
      "epoch": 1.73,
      "learning_rate": 7.828384292230813e-05,
      "loss": 1.3629,
      "step": 6765
    },
    {
      "epoch": 1.73,
      "learning_rate": 7.815260849927181e-05,
      "loss": 1.3513,
      "step": 6770
    },
    {
      "epoch": 1.73,
      "learning_rate": 7.802141357694312e-05,
      "loss": 1.3583,
      "step": 6775
    },
    {
      "epoch": 1.74,
      "learning_rate": 7.789025839252626e-05,
      "loss": 1.3875,
      "step": 6780
    },
    {
      "epoch": 1.74,
      "learning_rate": 7.775914318315356e-05,
      "loss": 1.3683,
      "step": 6785
    },
    {
      "epoch": 1.74,
      "learning_rate": 7.762806818588502e-05,
      "loss": 1.3166,
      "step": 6790
    },
    {
      "epoch": 1.74,
      "learning_rate": 7.749703363770797e-05,
      "loss": 1.3417,
      "step": 6795
    },
    {
      "epoch": 1.74,
      "learning_rate": 7.73660397755367e-05,
      "loss": 1.3855,
      "step": 6800
    },
    {
      "epoch": 1.74,
      "learning_rate": 7.723508683621173e-05,
      "loss": 1.3646,
      "step": 6805
    },
    {
      "epoch": 1.74,
      "learning_rate": 7.71041750564998e-05,
      "loss": 1.3429,
      "step": 6810
    },
    {
      "epoch": 1.74,
      "learning_rate": 7.697330467309315e-05,
      "loss": 1.3614,
      "step": 6815
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.684247592260911e-05,
      "loss": 1.322,
      "step": 6820
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.671168904158982e-05,
      "loss": 1.3567,
      "step": 6825
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.658094426650173e-05,
      "loss": 1.3313,
      "step": 6830
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.645024183373509e-05,
      "loss": 1.2917,
      "step": 6835
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.631958197960363e-05,
      "loss": 1.3379,
      "step": 6840
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.618896494034412e-05,
      "loss": 1.3761,
      "step": 6845
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.605839095211586e-05,
      "loss": 1.393,
      "step": 6850
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.592786025100039e-05,
      "loss": 1.3642,
      "step": 6855
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.579737307300092e-05,
      "loss": 1.3784,
      "step": 6860
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.566692965404205e-05,
      "loss": 1.3717,
      "step": 6865
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.55626065838749e-05,
      "loss": 1.3332,
      "step": 6870
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.543224252546602e-05,
      "loss": 1.3377,
      "step": 6875
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.532798324925699e-05,
      "loss": 1.3931,
      "step": 6880
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.519769931507215e-05,
      "loss": 1.3665,
      "step": 6885
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.506746022415514e-05,
      "loss": 1.3628,
      "step": 6890
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.493726621198199e-05,
      "loss": 1.3418,
      "step": 6895
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.480711751394712e-05,
      "loss": 1.3913,
      "step": 6900
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.467701436536312e-05,
      "loss": 1.3581,
      "step": 6905
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.454695700146024e-05,
      "loss": 1.3304,
      "step": 6910
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.441694565738587e-05,
      "loss": 1.3737,
      "step": 6915
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.42869805682042e-05,
      "loss": 1.3514,
      "step": 6920
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.415706196889592e-05,
      "loss": 1.3778,
      "step": 6925
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.402719009435745e-05,
      "loss": 1.348,
      "step": 6930
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.389736517940089e-05,
      "loss": 1.3409,
      "step": 6935
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.376758745875346e-05,
      "loss": 1.3892,
      "step": 6940
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.363785716705688e-05,
      "loss": 1.3793,
      "step": 6945
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.350817453886729e-05,
      "loss": 1.3446,
      "step": 6950
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.337853980865459e-05,
      "loss": 1.3599,
      "step": 6955
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.324895321080201e-05,
      "loss": 1.3812,
      "step": 6960
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.311941497960586e-05,
      "loss": 1.3655,
      "step": 6965
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.298992534927498e-05,
      "loss": 1.2942,
      "step": 6970
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.28604845539303e-05,
      "loss": 1.3597,
      "step": 6975
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.273109282760447e-05,
      "loss": 1.3486,
      "step": 6980
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.26017504042414e-05,
      "loss": 1.3881,
      "step": 6985
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.247245751769596e-05,
      "loss": 1.408,
      "step": 6990
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.23432144017333e-05,
      "loss": 1.312,
      "step": 6995
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.22398559008156e-05,
      "loss": 1.368,
      "step": 7000
    },
    {
      "epoch": 1.79,
      "eval_loss": 1.3932809829711914,
      "eval_runtime": 2.0256,
      "eval_samples_per_second": 49.369,
      "eval_steps_per_second": 1.975,
      "step": 7000
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.211070296070288e-05,
      "loss": 1.3496,
      "step": 7005
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.200741690315187e-05,
      "loss": 1.3066,
      "step": 7010
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.187835489546838e-05,
      "loss": 1.3885,
      "step": 7015
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.174934373252222e-05,
      "loss": 1.3138,
      "step": 7020
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.162038364756922e-05,
      "loss": 1.3625,
      "step": 7025
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.151725251245094e-05,
      "loss": 1.3357,
      "step": 7030
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.138838495539538e-05,
      "loss": 1.3793,
      "step": 7035
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.125956912895679e-05,
      "loss": 1.3629,
      "step": 7040
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.113080526603792e-05,
      "loss": 1.3643,
      "step": 7045
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.100209359944739e-05,
      "loss": 1.3606,
      "step": 7050
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.087343436189962e-05,
      "loss": 1.3523,
      "step": 7055
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.074482778601421e-05,
      "loss": 1.3189,
      "step": 7060
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.061627410431549e-05,
      "loss": 1.3774,
      "step": 7065
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.048777354923217e-05,
      "loss": 1.3047,
      "step": 7070
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.0359326353097e-05,
      "loss": 1.3835,
      "step": 7075
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.023093274814608e-05,
      "loss": 1.3136,
      "step": 7080
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.010259296651876e-05,
      "loss": 1.3299,
      "step": 7085
    },
    {
      "epoch": 1.81,
      "learning_rate": 6.997430724025706e-05,
      "loss": 1.3577,
      "step": 7090
    },
    {
      "epoch": 1.82,
      "learning_rate": 6.984607580130518e-05,
      "loss": 1.3134,
      "step": 7095
    },
    {
      "epoch": 1.82,
      "learning_rate": 6.971789888150926e-05,
      "loss": 1.3005,
      "step": 7100
    },
    {
      "epoch": 1.82,
      "learning_rate": 6.958977671261683e-05,
      "loss": 1.3304,
      "step": 7105
    },
    {
      "epoch": 1.82,
      "learning_rate": 6.94617095262764e-05,
      "loss": 1.4018,
      "step": 7110
    },
    {
      "epoch": 1.82,
      "learning_rate": 6.933369755403711e-05,
      "loss": 1.285,
      "step": 7115
    },
    {
      "epoch": 1.82,
      "learning_rate": 6.920574102734827e-05,
      "loss": 1.3071,
      "step": 7120
    },
    {
      "epoch": 1.82,
      "learning_rate": 6.907784017755896e-05,
      "loss": 1.3059,
      "step": 7125
    },
    {
      "epoch": 1.83,
      "learning_rate": 6.894999523591752e-05,
      "loss": 1.3412,
      "step": 7130
    },
    {
      "epoch": 1.83,
      "learning_rate": 6.882220643357129e-05,
      "loss": 1.3371,
      "step": 7135
    },
    {
      "epoch": 1.83,
      "learning_rate": 6.869447400156608e-05,
      "loss": 1.3368,
      "step": 7140
    },
    {
      "epoch": 1.83,
      "learning_rate": 6.856679817084571e-05,
      "loss": 1.3812,
      "step": 7145
    },
    {
      "epoch": 1.83,
      "learning_rate": 6.843917917225178e-05,
      "loss": 1.3641,
      "step": 7150
    },
    {
      "epoch": 1.83,
      "learning_rate": 6.831161723652313e-05,
      "loss": 1.3325,
      "step": 7155
    },
    {
      "epoch": 1.83,
      "learning_rate": 6.818411259429529e-05,
      "loss": 1.3737,
      "step": 7160
    },
    {
      "epoch": 1.83,
      "learning_rate": 6.80566654761003e-05,
      "loss": 1.3421,
      "step": 7165
    },
    {
      "epoch": 1.84,
      "learning_rate": 6.792927611236628e-05,
      "loss": 1.3544,
      "step": 7170
    },
    {
      "epoch": 1.84,
      "learning_rate": 6.780194473341675e-05,
      "loss": 1.3373,
      "step": 7175
    },
    {
      "epoch": 1.84,
      "learning_rate": 6.767467156947049e-05,
      "loss": 1.3546,
      "step": 7180
    },
    {
      "epoch": 1.84,
      "learning_rate": 6.754745685064105e-05,
      "loss": 1.342,
      "step": 7185
    },
    {
      "epoch": 1.84,
      "learning_rate": 6.742030080693625e-05,
      "loss": 1.3306,
      "step": 7190
    },
    {
      "epoch": 1.84,
      "learning_rate": 6.729320366825784e-05,
      "loss": 1.3223,
      "step": 7195
    },
    {
      "epoch": 1.84,
      "learning_rate": 6.716616566440108e-05,
      "loss": 1.3691,
      "step": 7200
    },
    {
      "epoch": 1.84,
      "learning_rate": 6.703918702505439e-05,
      "loss": 1.3439,
      "step": 7205
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.691226797979864e-05,
      "loss": 1.3378,
      "step": 7210
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.678540875810712e-05,
      "loss": 1.3042,
      "step": 7215
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.665860958934504e-05,
      "loss": 1.32,
      "step": 7220
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.653187070276877e-05,
      "loss": 1.3102,
      "step": 7225
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.64051923275259e-05,
      "loss": 1.305,
      "step": 7230
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.62785746926546e-05,
      "loss": 1.3073,
      "step": 7235
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.615201802708307e-05,
      "loss": 1.3606,
      "step": 7240
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.602552255962942e-05,
      "loss": 1.3425,
      "step": 7245
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.589908851900112e-05,
      "loss": 1.3135,
      "step": 7250
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.577271613379442e-05,
      "loss": 1.3483,
      "step": 7255
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.564640563249426e-05,
      "loss": 1.3177,
      "step": 7260
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.552015724347366e-05,
      "loss": 1.3023,
      "step": 7265
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.539397119499326e-05,
      "loss": 1.2986,
      "step": 7270
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.526784771520107e-05,
      "loss": 1.357,
      "step": 7275
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.514178703213192e-05,
      "loss": 1.3706,
      "step": 7280
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.501578937370716e-05,
      "loss": 1.3431,
      "step": 7285
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.48898549677341e-05,
      "loss": 1.3846,
      "step": 7290
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.476398404190578e-05,
      "loss": 1.3383,
      "step": 7295
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.463817682380045e-05,
      "loss": 1.3783,
      "step": 7300
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.451243354088107e-05,
      "loss": 1.3338,
      "step": 7305
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.438675442049518e-05,
      "loss": 1.321,
      "step": 7310
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.42611396898742e-05,
      "loss": 1.3351,
      "step": 7315
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.413558957613308e-05,
      "loss": 1.331,
      "step": 7320
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.401010430627006e-05,
      "loss": 1.3132,
      "step": 7325
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.388468410716613e-05,
      "loss": 1.3398,
      "step": 7330
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.380946331600652e-05,
      "loss": 1.2932,
      "step": 7335
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.370920557903063e-05,
      "loss": 1.3546,
      "step": 7340
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.358394247471778e-05,
      "loss": 1.326,
      "step": 7345
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.345874521167675e-05,
      "loss": 1.3047,
      "step": 7350
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.333361401626776e-05,
      "loss": 1.3499,
      "step": 7355
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.320854911473157e-05,
      "loss": 1.3802,
      "step": 7360
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.308355073318911e-05,
      "loss": 1.3391,
      "step": 7365
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.295861909764105e-05,
      "loss": 1.3309,
      "step": 7370
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.283375443396726e-05,
      "loss": 1.3428,
      "step": 7375
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.270895696792669e-05,
      "loss": 1.3402,
      "step": 7380
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.258422692515673e-05,
      "loss": 1.3278,
      "step": 7385
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.245956453117282e-05,
      "loss": 1.3833,
      "step": 7390
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.233497001136815e-05,
      "loss": 1.3758,
      "step": 7395
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.221044359101317e-05,
      "loss": 1.3332,
      "step": 7400
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.208598549525518e-05,
      "loss": 1.367,
      "step": 7405
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.196159594911799e-05,
      "loss": 1.3435,
      "step": 7410
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.18372751775015e-05,
      "loss": 1.3399,
      "step": 7415
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.171302340518111e-05,
      "loss": 1.3217,
      "step": 7420
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.158884085680762e-05,
      "loss": 1.3208,
      "step": 7425
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.146472775690665e-05,
      "loss": 1.3457,
      "step": 7430
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.134068432987825e-05,
      "loss": 1.3435,
      "step": 7435
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.121671079999638e-05,
      "loss": 1.2834,
      "step": 7440
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.10928073914088e-05,
      "loss": 1.3118,
      "step": 7445
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.09689743281364e-05,
      "loss": 1.2927,
      "step": 7450
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.084521183407286e-05,
      "loss": 1.3626,
      "step": 7455
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.0721520132984335e-05,
      "loss": 1.3948,
      "step": 7460
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.0597899448508955e-05,
      "loss": 1.299,
      "step": 7465
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.047435000415642e-05,
      "loss": 1.3198,
      "step": 7470
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.035087202330767e-05,
      "loss": 1.3511,
      "step": 7475
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.022746572921447e-05,
      "loss": 1.3859,
      "step": 7480
    },
    {
      "epoch": 1.92,
      "learning_rate": 6.0104131344998814e-05,
      "loss": 1.3216,
      "step": 7485
    },
    {
      "epoch": 1.92,
      "learning_rate": 6.003016532398079e-05,
      "loss": 1.3332,
      "step": 7490
    },
    {
      "epoch": 1.92,
      "learning_rate": 5.9906946459338656e-05,
      "loss": 1.307,
      "step": 7495
    },
    {
      "epoch": 1.92,
      "learning_rate": 5.978380008408209e-05,
      "loss": 1.3368,
      "step": 7500
    },
    {
      "epoch": 1.92,
      "learning_rate": 5.9660726420863255e-05,
      "loss": 1.3286,
      "step": 7505
    },
    {
      "epoch": 1.92,
      "learning_rate": 5.9537725692202906e-05,
      "loss": 1.3934,
      "step": 7510
    },
    {
      "epoch": 1.92,
      "learning_rate": 5.941479812048982e-05,
      "loss": 1.3065,
      "step": 7515
    },
    {
      "epoch": 1.92,
      "learning_rate": 5.9291943927980584e-05,
      "loss": 1.2723,
      "step": 7520
    },
    {
      "epoch": 1.93,
      "learning_rate": 5.9169163336799085e-05,
      "loss": 1.3582,
      "step": 7525
    },
    {
      "epoch": 1.93,
      "learning_rate": 5.904645656893615e-05,
      "loss": 1.3274,
      "step": 7530
    },
    {
      "epoch": 1.93,
      "learning_rate": 5.892382384624907e-05,
      "loss": 1.3052,
      "step": 7535
    },
    {
      "epoch": 1.93,
      "learning_rate": 5.880126539046136e-05,
      "loss": 1.359,
      "step": 7540
    },
    {
      "epoch": 1.93,
      "learning_rate": 5.867878142316221e-05,
      "loss": 1.2711,
      "step": 7545
    },
    {
      "epoch": 1.93,
      "learning_rate": 5.855637216580605e-05,
      "loss": 1.3067,
      "step": 7550
    },
    {
      "epoch": 1.93,
      "learning_rate": 5.84340378397124e-05,
      "loss": 1.3473,
      "step": 7555
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.831177866606523e-05,
      "loss": 1.3167,
      "step": 7560
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.81895948659125e-05,
      "loss": 1.2768,
      "step": 7565
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.806748666016612e-05,
      "loss": 1.3302,
      "step": 7570
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.7945454269601185e-05,
      "loss": 1.298,
      "step": 7575
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.782349791485573e-05,
      "loss": 1.2749,
      "step": 7580
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.7725987725030394e-05,
      "loss": 1.296,
      "step": 7585
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.760416879032877e-05,
      "loss": 1.3993,
      "step": 7590
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.748242650850055e-05,
      "loss": 1.3955,
      "step": 7595
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.7360761099659235e-05,
      "loss": 1.3511,
      "step": 7600
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.723917278377931e-05,
      "loss": 1.3161,
      "step": 7605
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.711766178069597e-05,
      "loss": 1.3537,
      "step": 7610
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.69962283101046e-05,
      "loss": 1.2801,
      "step": 7615
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.6874872591560346e-05,
      "loss": 1.3439,
      "step": 7620
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.675359484447776e-05,
      "loss": 1.2978,
      "step": 7625
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.663239528813058e-05,
      "loss": 1.2987,
      "step": 7630
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.6511274141651005e-05,
      "loss": 1.3354,
      "step": 7635
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.63902316240295e-05,
      "loss": 1.2766,
      "step": 7640
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.626926795411447e-05,
      "loss": 1.3342,
      "step": 7645
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.6148383350611636e-05,
      "loss": 1.3392,
      "step": 7650
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.602757803208377e-05,
      "loss": 1.2999,
      "step": 7655
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.5906852216950434e-05,
      "loss": 1.3485,
      "step": 7660
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.5786206123487295e-05,
      "loss": 1.3972,
      "step": 7665
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.56656399698259e-05,
      "loss": 1.3637,
      "step": 7670
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.5545153973953377e-05,
      "loss": 1.312,
      "step": 7675
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.542474835371181e-05,
      "loss": 1.3149,
      "step": 7680
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.530442332679799e-05,
      "loss": 1.2972,
      "step": 7685
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.518417911076297e-05,
      "loss": 1.2729,
      "step": 7690
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.5064015923011804e-05,
      "loss": 1.3449,
      "step": 7695
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.494393398080292e-05,
      "loss": 1.3416,
      "step": 7700
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.482393350124786e-05,
      "loss": 1.3982,
      "step": 7705
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.4704014701311013e-05,
      "loss": 1.3364,
      "step": 7710
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.458417779780893e-05,
      "loss": 1.2942,
      "step": 7715
    },
    {
      "epoch": 1.98,
      "learning_rate": 5.4464423007410125e-05,
      "loss": 1.3555,
      "step": 7720
    },
    {
      "epoch": 1.98,
      "learning_rate": 5.4344750546634805e-05,
      "loss": 1.3627,
      "step": 7725
    },
    {
      "epoch": 1.98,
      "learning_rate": 5.422516063185403e-05,
      "loss": 1.2841,
      "step": 7730
    },
    {
      "epoch": 1.98,
      "learning_rate": 5.41056534792899e-05,
      "loss": 1.3398,
      "step": 7735
    },
    {
      "epoch": 1.98,
      "learning_rate": 5.398622930501467e-05,
      "loss": 1.3023,
      "step": 7740
    },
    {
      "epoch": 1.98,
      "learning_rate": 5.386688832495076e-05,
      "loss": 1.3116,
      "step": 7745
    },
    {
      "epoch": 1.98,
      "learning_rate": 5.374763075487002e-05,
      "loss": 1.336,
      "step": 7750
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.3628456810393476e-05,
      "loss": 1.3207,
      "step": 7755
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.350936670699114e-05,
      "loss": 1.3143,
      "step": 7760
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.3390360659981266e-05,
      "loss": 1.3169,
      "step": 7765
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.327143888453013e-05,
      "loss": 1.3261,
      "step": 7770
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.3152601595651843e-05,
      "loss": 1.3181,
      "step": 7775
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.303384900820744e-05,
      "loss": 1.3681,
      "step": 7780
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.291518133690507e-05,
      "loss": 1.2813,
      "step": 7785
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.2796598796299356e-05,
      "loss": 1.3234,
      "step": 7790
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.267810160079076e-05,
      "loss": 1.3349,
      "step": 7795
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.255968996462571e-05,
      "loss": 1.331,
      "step": 7800
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.24413641018958e-05,
      "loss": 1.3721,
      "step": 7805
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.232312422653749e-05,
      "loss": 1.3645,
      "step": 7810
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.2228594380824624e-05,
      "loss": 1.1685,
      "step": 7815
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.21105098213589e-05,
      "loss": 0.9281,
      "step": 7820
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.199251184745883e-05,
      "loss": 0.8604,
      "step": 7825
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.189817595331834e-05,
      "loss": 0.7993,
      "step": 7830
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.1780334370957174e-05,
      "loss": 0.8365,
      "step": 7835
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.166257997112825e-05,
      "loss": 0.8685,
      "step": 7840
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.1544912966734994e-05,
      "loss": 0.8127,
      "step": 7845
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.1427333570522604e-05,
      "loss": 0.8241,
      "step": 7850
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.130984199507811e-05,
      "loss": 0.8348,
      "step": 7855
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.1192438452829484e-05,
      "loss": 0.8307,
      "step": 7860
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.10751231560458e-05,
      "loss": 0.8332,
      "step": 7865
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.098133459789646e-05,
      "loss": 0.7993,
      "step": 7870
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.086417867735782e-05,
      "loss": 0.8072,
      "step": 7875
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.074711159578711e-05,
      "loss": 0.8111,
      "step": 7880
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.0630133564845004e-05,
      "loss": 0.8453,
      "step": 7885
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.051324479603106e-05,
      "loss": 0.8498,
      "step": 7890
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.039644550068361e-05,
      "loss": 0.82,
      "step": 7895
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.027973588997896e-05,
      "loss": 0.8221,
      "step": 7900
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.0163116174931546e-05,
      "loss": 0.8248,
      "step": 7905
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.004658656639311e-05,
      "loss": 0.7862,
      "step": 7910
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.9930147275052455e-05,
      "loss": 0.7912,
      "step": 7915
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.9813798511435236e-05,
      "loss": 0.826,
      "step": 7920
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.9697540485903295e-05,
      "loss": 0.8136,
      "step": 7925
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.958137340865442e-05,
      "loss": 0.8145,
      "step": 7930
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.946529748972209e-05,
      "loss": 0.8301,
      "step": 7935
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.9349312938974835e-05,
      "loss": 0.8612,
      "step": 7940
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.9233419966116036e-05,
      "loss": 0.8155,
      "step": 7945
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.911761878068345e-05,
      "loss": 0.8126,
      "step": 7950
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.900190959204902e-05,
      "loss": 0.8259,
      "step": 7955
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.888629260941823e-05,
      "loss": 0.837,
      "step": 7960
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.877076804182982e-05,
      "loss": 0.8169,
      "step": 7965
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.86553360981556e-05,
      "loss": 0.8136,
      "step": 7970
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.853999698709978e-05,
      "loss": 0.8307,
      "step": 7975
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.8424750917198725e-05,
      "loss": 0.8249,
      "step": 7980
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.830959809682071e-05,
      "loss": 0.8184,
      "step": 7985
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.819453873416526e-05,
      "loss": 0.8315,
      "step": 7990
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.807957303726297e-05,
      "loss": 0.7969,
      "step": 7995
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.7964701213975174e-05,
      "loss": 0.8141,
      "step": 8000
    },
    {
      "epoch": 2.05,
      "eval_loss": 1.541500210762024,
      "eval_runtime": 2.0224,
      "eval_samples_per_second": 49.447,
      "eval_steps_per_second": 1.978,
      "step": 8000
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.784992347199336e-05,
      "loss": 0.7866,
      "step": 8005
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.773524001883898e-05,
      "loss": 0.8183,
      "step": 8010
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.7620651061862945e-05,
      "loss": 0.8114,
      "step": 8015
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.750615680824543e-05,
      "loss": 0.8187,
      "step": 8020
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.739175746499528e-05,
      "loss": 0.8116,
      "step": 8025
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.727745323894976e-05,
      "loss": 0.8192,
      "step": 8030
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.716324433677422e-05,
      "loss": 0.7999,
      "step": 8035
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.7049130964961575e-05,
      "loss": 0.822,
      "step": 8040
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.693511332983205e-05,
      "loss": 0.8005,
      "step": 8045
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.682119163753288e-05,
      "loss": 0.8102,
      "step": 8050
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.67073660940376e-05,
      "loss": 0.8282,
      "step": 8055
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.659363690514613e-05,
      "loss": 0.8053,
      "step": 8060
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.648000427648405e-05,
      "loss": 0.7925,
      "step": 8065
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.636646841350244e-05,
      "loss": 0.7961,
      "step": 8070
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.6253029521477364e-05,
      "loss": 0.8349,
      "step": 8075
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.616234836477959e-05,
      "loss": 0.791,
      "step": 8080
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.6049084537212075e-05,
      "loss": 0.7877,
      "step": 8085
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.593591825444028e-05,
      "loss": 0.8458,
      "step": 8090
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.5822849721071916e-05,
      "loss": 0.8374,
      "step": 8095
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.570987914153823e-05,
      "loss": 0.8018,
      "step": 8100
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.5597006720093195e-05,
      "loss": 0.7831,
      "step": 8105
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.548423266081344e-05,
      "loss": 0.8071,
      "step": 8110
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.537155716759763e-05,
      "loss": 0.8404,
      "step": 8115
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.5258980444166276e-05,
      "loss": 0.7918,
      "step": 8120
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.514650269406134e-05,
      "loss": 0.7938,
      "step": 8125
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.503412412064579e-05,
      "loss": 0.8247,
      "step": 8130
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.4921844927103264e-05,
      "loss": 0.8304,
      "step": 8135
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.483209326220262e-05,
      "loss": 0.8135,
      "step": 8140
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.47199934638807e-05,
      "loss": 0.7624,
      "step": 8145
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.4607993613388976e-05,
      "loss": 0.8426,
      "step": 8150
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.4496093913226475e-05,
      "loss": 0.8086,
      "step": 8155
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.438429456571097e-05,
      "loss": 0.8011,
      "step": 8160
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.4272595772978944e-05,
      "loss": 0.7947,
      "step": 8165
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.416099773698497e-05,
      "loss": 0.7717,
      "step": 8170
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.4049500659501444e-05,
      "loss": 0.789,
      "step": 8175
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.393810474211837e-05,
      "loss": 0.7948,
      "step": 8180
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.3826810186242615e-05,
      "loss": 0.827,
      "step": 8185
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.371561719309798e-05,
      "loss": 0.8249,
      "step": 8190
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.360452596372464e-05,
      "loss": 0.7775,
      "step": 8195
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.3493536698978566e-05,
      "loss": 0.7961,
      "step": 8200
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.3382649599531636e-05,
      "loss": 0.8309,
      "step": 8205
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.327186486587085e-05,
      "loss": 0.8117,
      "step": 8210
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.316118269829812e-05,
      "loss": 0.8262,
      "step": 8215
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.307271094630797e-05,
      "loss": 0.8273,
      "step": 8220
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.296221390186261e-05,
      "loss": 0.8161,
      "step": 8225
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.2851819983363185e-05,
      "loss": 0.7935,
      "step": 8230
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.2741529390405e-05,
      "loss": 0.7901,
      "step": 8235
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.263134232239657e-05,
      "loss": 0.8162,
      "step": 8240
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.252125897855932e-05,
      "loss": 0.7729,
      "step": 8245
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.2411279557927064e-05,
      "loss": 0.8068,
      "step": 8250
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.2301404259345665e-05,
      "loss": 0.8164,
      "step": 8255
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.2191633281472905e-05,
      "loss": 0.7805,
      "step": 8260
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.208196682277781e-05,
      "loss": 0.8267,
      "step": 8265
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.1972405081540445e-05,
      "loss": 0.7728,
      "step": 8270
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.186294825585154e-05,
      "loss": 0.789,
      "step": 8275
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.1753596543612236e-05,
      "loss": 0.8235,
      "step": 8280
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.1644350142533514e-05,
      "loss": 0.8251,
      "step": 8285
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.1535209250135964e-05,
      "loss": 0.7859,
      "step": 8290
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.142617406374948e-05,
      "loss": 0.7764,
      "step": 8295
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.131724478051278e-05,
      "loss": 0.7557,
      "step": 8300
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.120842159737307e-05,
      "loss": 0.817,
      "step": 8305
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.109970471108585e-05,
      "loss": 0.8062,
      "step": 8310
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.0991094318214305e-05,
      "loss": 0.7775,
      "step": 8315
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.088259061512912e-05,
      "loss": 0.8345,
      "step": 8320
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.077419379800813e-05,
      "loss": 0.78,
      "step": 8325
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.0665904062835856e-05,
      "loss": 0.79,
      "step": 8330
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.055772160540323e-05,
      "loss": 0.8021,
      "step": 8335
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.044964662130719e-05,
      "loss": 0.8264,
      "step": 8340
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.036326414614985e-05,
      "loss": 0.8133,
      "step": 8345
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.0255383106343646e-05,
      "loss": 0.7864,
      "step": 8350
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.014761008651069e-05,
      "loss": 0.7778,
      "step": 8355
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.0039945281507665e-05,
      "loss": 0.844,
      "step": 8360
    },
    {
      "epoch": 2.14,
      "learning_rate": 3.9932388885995654e-05,
      "loss": 0.7857,
      "step": 8365
    },
    {
      "epoch": 2.14,
      "learning_rate": 3.9824941094439685e-05,
      "loss": 0.7739,
      "step": 8370
    },
    {
      "epoch": 2.14,
      "learning_rate": 3.971760210110841e-05,
      "loss": 0.7892,
      "step": 8375
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.961037210007378e-05,
      "loss": 0.8123,
      "step": 8380
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.950325128521075e-05,
      "loss": 0.7909,
      "step": 8385
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.939623985019679e-05,
      "loss": 0.7824,
      "step": 8390
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.9289337988511584e-05,
      "loss": 0.788,
      "step": 8395
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.918254589343683e-05,
      "loss": 0.8426,
      "step": 8400
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.9075863758055655e-05,
      "loss": 0.7851,
      "step": 8405
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.896929177525235e-05,
      "loss": 0.7971,
      "step": 8410
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.886283013771223e-05,
      "loss": 0.7831,
      "step": 8415
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.8756479037920815e-05,
      "loss": 0.7948,
      "step": 8420
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.865023866816405e-05,
      "loss": 0.7827,
      "step": 8425
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.854410922052748e-05,
      "loss": 0.7868,
      "step": 8430
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.843809088689626e-05,
      "loss": 0.8085,
      "step": 8435
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.833218385895451e-05,
      "loss": 0.7999,
      "step": 8440
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.8226388328185136e-05,
      "loss": 0.7836,
      "step": 8445
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.812070448586955e-05,
      "loss": 0.8044,
      "step": 8450
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.8015132523087116e-05,
      "loss": 0.8189,
      "step": 8455
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.790967263071492e-05,
      "loss": 0.796,
      "step": 8460
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.7825385535652703e-05,
      "loss": 0.7954,
      "step": 8465
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.772012785038236e-05,
      "loss": 0.7662,
      "step": 8470
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.76149827688992e-05,
      "loss": 0.7668,
      "step": 8475
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.7509950481308496e-05,
      "loss": 0.8148,
      "step": 8480
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.7405031177511594e-05,
      "loss": 0.8042,
      "step": 8485
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.730022504720566e-05,
      "loss": 0.7851,
      "step": 8490
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.721646175522023e-05,
      "loss": 0.8073,
      "step": 8495
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.711185981458125e-05,
      "loss": 0.7768,
      "step": 8500
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.700737157749543e-05,
      "loss": 0.7647,
      "step": 8505
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.6902997232880454e-05,
      "loss": 0.7943,
      "step": 8510
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.679873696944809e-05,
      "loss": 0.8003,
      "step": 8515
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.671541102383639e-05,
      "loss": 0.7862,
      "step": 8520
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.6611356581427884e-05,
      "loss": 0.7788,
      "step": 8525
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.6507416747496634e-05,
      "loss": 0.8285,
      "step": 8530
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.640359170996892e-05,
      "loss": 0.8049,
      "step": 8535
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.6299881656563206e-05,
      "loss": 0.8017,
      "step": 8540
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.6196286774790354e-05,
      "loss": 0.8083,
      "step": 8545
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.6092807251952824e-05,
      "loss": 0.8056,
      "step": 8550
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.598944327514452e-05,
      "loss": 0.7709,
      "step": 8555
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.588619503125052e-05,
      "loss": 0.8217,
      "step": 8560
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.578306270694658e-05,
      "loss": 0.7695,
      "step": 8565
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.568004648869884e-05,
      "loss": 0.7713,
      "step": 8570
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.557714656276363e-05,
      "loss": 0.7982,
      "step": 8575
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.5474363115186905e-05,
      "loss": 0.7633,
      "step": 8580
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.537169633180416e-05,
      "loss": 0.799,
      "step": 8585
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.5269146398239726e-05,
      "loss": 0.8049,
      "step": 8590
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.516671349990693e-05,
      "loss": 0.8212,
      "step": 8595
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.506439782200735e-05,
      "loss": 0.778,
      "step": 8600
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.496219954953061e-05,
      "loss": 0.8048,
      "step": 8605
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.486011886725414e-05,
      "loss": 0.7691,
      "step": 8610
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.4758155959742755e-05,
      "loss": 0.8099,
      "step": 8615
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.465631101134822e-05,
      "loss": 0.8097,
      "step": 8620
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.455458420620921e-05,
      "loss": 0.8042,
      "step": 8625
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.4452975728250626e-05,
      "loss": 0.7527,
      "step": 8630
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.435148576118353e-05,
      "loss": 0.763,
      "step": 8635
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.425011448850463e-05,
      "loss": 0.8004,
      "step": 8640
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.414886209349615e-05,
      "loss": 0.7536,
      "step": 8645
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.404772875922528e-05,
      "loss": 0.7591,
      "step": 8650
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.3946714668543945e-05,
      "loss": 0.7937,
      "step": 8655
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.384582000408857e-05,
      "loss": 0.8083,
      "step": 8660
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.3745044948279544e-05,
      "loss": 0.7987,
      "step": 8665
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.3644389683321044e-05,
      "loss": 0.787,
      "step": 8670
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.354385439120069e-05,
      "loss": 0.7325,
      "step": 8675
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.344343925368916e-05,
      "loss": 0.802,
      "step": 8680
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.3343144452339836e-05,
      "loss": 0.7558,
      "step": 8685
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.3242970168488654e-05,
      "loss": 0.7981,
      "step": 8690
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.3142916583253546e-05,
      "loss": 0.7896,
      "step": 8695
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.304298387753426e-05,
      "loss": 0.7875,
      "step": 8700
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.294317223201194e-05,
      "loss": 0.7928,
      "step": 8705
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.284348182714896e-05,
      "loss": 0.815,
      "step": 8710
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.274391284318838e-05,
      "loss": 0.784,
      "step": 8715
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.264446546015375e-05,
      "loss": 0.8006,
      "step": 8720
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.2545139857848814e-05,
      "loss": 0.7632,
      "step": 8725
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.2445936215857096e-05,
      "loss": 0.7857,
      "step": 8730
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.234685471354155e-05,
      "loss": 0.7726,
      "step": 8735
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.224789553004446e-05,
      "loss": 0.7899,
      "step": 8740
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.21490588442868e-05,
      "loss": 0.7997,
      "step": 8745
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.2070077814146595e-05,
      "loss": 0.8176,
      "step": 8750
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.199117538832358e-05,
      "loss": 0.8142,
      "step": 8755
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.18926580324486e-05,
      "loss": 0.7462,
      "step": 8760
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.179426381659256e-05,
      "loss": 0.7717,
      "step": 8765
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.169599291865509e-05,
      "loss": 0.8044,
      "step": 8770
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.159784551631273e-05,
      "loss": 0.7837,
      "step": 8775
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.149982178701879e-05,
      "loss": 0.7939,
      "step": 8780
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.140192190800294e-05,
      "loss": 0.7561,
      "step": 8785
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.1304146056271024e-05,
      "loss": 0.801,
      "step": 8790
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.120649440860454e-05,
      "loss": 0.7872,
      "step": 8795
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.110896714156042e-05,
      "loss": 0.7874,
      "step": 8800
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.101156443147082e-05,
      "loss": 0.7947,
      "step": 8805
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.091428645444257e-05,
      "loss": 0.7554,
      "step": 8810
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.081713338635702e-05,
      "loss": 0.8199,
      "step": 8815
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.072010540286978e-05,
      "loss": 0.8038,
      "step": 8820
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.062320267941008e-05,
      "loss": 0.7471,
      "step": 8825
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.0526425391180925e-05,
      "loss": 0.7925,
      "step": 8830
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.0429773713158327e-05,
      "loss": 0.7586,
      "step": 8835
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.0333247820091348e-05,
      "loss": 0.8077,
      "step": 8840
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.0236847886501542e-05,
      "loss": 0.7621,
      "step": 8845
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.0140574086682705e-05,
      "loss": 0.7724,
      "step": 8850
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.004442659470067e-05,
      "loss": 0.762,
      "step": 8855
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.9967599659583968e-05,
      "loss": 0.7707,
      "step": 8860
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.987167995962471e-05,
      "loss": 0.7806,
      "step": 8865
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.9775887053670513e-05,
      "loss": 0.7871,
      "step": 8870
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.9699344136986406e-05,
      "loss": 0.8071,
      "step": 8875
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.9603779896541707e-05,
      "loss": 0.7619,
      "step": 8880
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.9508342934473054e-05,
      "loss": 0.7615,
      "step": 8885
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.9413033423333092e-05,
      "loss": 0.8218,
      "step": 8890
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.931785153544421e-05,
      "loss": 0.7685,
      "step": 8895
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.922279744289802e-05,
      "loss": 0.7681,
      "step": 8900
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.9127871317554922e-05,
      "loss": 0.77,
      "step": 8905
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.903307333104416e-05,
      "loss": 0.763,
      "step": 8910
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.893840365476316e-05,
      "loss": 0.7474,
      "step": 8915
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.8843862459877368e-05,
      "loss": 0.7853,
      "step": 8920
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.8749449917320036e-05,
      "loss": 0.8067,
      "step": 8925
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.865516619779166e-05,
      "loss": 0.8306,
      "step": 8930
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.856101147175998e-05,
      "loss": 0.7622,
      "step": 8935
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.8466985909459376e-05,
      "loss": 0.7536,
      "step": 8940
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.8373089680890743e-05,
      "loss": 0.7628,
      "step": 8945
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.827932295582121e-05,
      "loss": 0.764,
      "step": 8950
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.8185685903783578e-05,
      "loss": 0.801,
      "step": 8955
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.809217869407641e-05,
      "loss": 0.7734,
      "step": 8960
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.7998801495763383e-05,
      "loss": 0.8021,
      "step": 8965
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.790555447767309e-05,
      "loss": 0.7698,
      "step": 8970
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.7812437808398883e-05,
      "loss": 0.7555,
      "step": 8975
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.7719451656298313e-05,
      "loss": 0.7925,
      "step": 8980
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.7626596189492983e-05,
      "loss": 0.7821,
      "step": 8985
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.7533871575868276e-05,
      "loss": 0.7577,
      "step": 8990
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.744127798307291e-05,
      "loss": 0.7862,
      "step": 8995
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.7348815578518762e-05,
      "loss": 0.7488,
      "step": 9000
    },
    {
      "epoch": 2.3,
      "eval_loss": 1.5792468786239624,
      "eval_runtime": 2.0258,
      "eval_samples_per_second": 49.363,
      "eval_steps_per_second": 1.975,
      "step": 9000
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.7256484529380442e-05,
      "loss": 0.7489,
      "step": 9005
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.7164285002595213e-05,
      "loss": 0.7806,
      "step": 9010
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.70722171648624e-05,
      "loss": 0.7767,
      "step": 9015
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.6980281182643252e-05,
      "loss": 0.7674,
      "step": 9020
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.6888477222160735e-05,
      "loss": 0.7475,
      "step": 9025
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.6796805449398976e-05,
      "loss": 0.7891,
      "step": 9030
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.6705266030103128e-05,
      "loss": 0.779,
      "step": 9035
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.663212990038998e-05,
      "loss": 0.7869,
      "step": 9040
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.654082913424668e-05,
      "loss": 0.7436,
      "step": 9045
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.6449661184381836e-05,
      "loss": 0.8064,
      "step": 9050
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.635862621562969e-05,
      "loss": 0.8043,
      "step": 9055
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.6267724392584014e-05,
      "loss": 0.806,
      "step": 9060
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.6176955879597987e-05,
      "loss": 0.7529,
      "step": 9065
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.608632084078363e-05,
      "loss": 0.795,
      "step": 9070
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.599581944001165e-05,
      "loss": 0.7276,
      "step": 9075
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.590545184091122e-05,
      "loss": 0.7709,
      "step": 9080
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.5815218206869517e-05,
      "loss": 0.792,
      "step": 9085
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.5725118701031492e-05,
      "loss": 0.7417,
      "step": 9090
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.563515348629969e-05,
      "loss": 0.7593,
      "step": 9095
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.5545322725333755e-05,
      "loss": 0.7706,
      "step": 9100
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.545562658055026e-05,
      "loss": 0.8017,
      "step": 9105
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.5366065214122392e-05,
      "loss": 0.7551,
      "step": 9110
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.527663878797968e-05,
      "loss": 0.7845,
      "step": 9115
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.5187347463807653e-05,
      "loss": 0.775,
      "step": 9120
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.5098191403047545e-05,
      "loss": 0.7423,
      "step": 9125
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.5009170766896108e-05,
      "loss": 0.7769,
      "step": 9130
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.4938051871861045e-05,
      "loss": 0.7546,
      "step": 9135
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.484927540543911e-05,
      "loss": 0.7554,
      "step": 9140
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.4760634813673057e-05,
      "loss": 0.7791,
      "step": 9145
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.4672130256827565e-05,
      "loss": 0.7771,
      "step": 9150
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.4583761894921463e-05,
      "loss": 0.7658,
      "step": 9155
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.4495529887727255e-05,
      "loss": 0.7632,
      "step": 9160
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.4407434394770878e-05,
      "loss": 0.8075,
      "step": 9165
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.4319475575331586e-05,
      "loss": 0.7777,
      "step": 9170
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.4231653588441383e-05,
      "loss": 0.7945,
      "step": 9175
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.4143968592884903e-05,
      "loss": 0.7736,
      "step": 9180
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.4056420747199215e-05,
      "loss": 0.7724,
      "step": 9185
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.3969010209673215e-05,
      "loss": 0.7701,
      "step": 9190
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.3881737138347728e-05,
      "loss": 0.7586,
      "step": 9195
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.3794601691014896e-05,
      "loss": 0.7825,
      "step": 9200
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.3707604025218154e-05,
      "loss": 0.7607,
      "step": 9205
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.3620744298251752e-05,
      "loss": 0.7596,
      "step": 9210
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.3534022667160516e-05,
      "loss": 0.7408,
      "step": 9215
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.3447439288739693e-05,
      "loss": 0.7688,
      "step": 9220
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.336099431953448e-05,
      "loss": 0.7484,
      "step": 9225
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.32746879158398e-05,
      "loss": 0.7823,
      "step": 9230
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.3205742664922004e-05,
      "loss": 0.7849,
      "step": 9235
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.3119686072209857e-05,
      "loss": 0.7696,
      "step": 9240
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.3033768481300443e-05,
      "loss": 0.7686,
      "step": 9245
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.2947990047535307e-05,
      "loss": 0.7633,
      "step": 9250
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.2862350926004227e-05,
      "loss": 0.7678,
      "step": 9255
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.2776851271545252e-05,
      "loss": 0.7738,
      "step": 9260
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.2708552068160115e-05,
      "loss": 0.7717,
      "step": 9265
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.2623303843817457e-05,
      "loss": 0.7829,
      "step": 9270
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.2538195518750848e-05,
      "loss": 0.7788,
      "step": 9275
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.2453227246838538e-05,
      "loss": 0.8016,
      "step": 9280
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.2368399181705567e-05,
      "loss": 0.835,
      "step": 9285
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.2283711476723544e-05,
      "loss": 0.729,
      "step": 9290
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.219916428501021e-05,
      "loss": 0.7551,
      "step": 9295
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.211475775942936e-05,
      "loss": 0.7486,
      "step": 9300
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.2030492052590368e-05,
      "loss": 0.7457,
      "step": 9305
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.1946367316848016e-05,
      "loss": 0.7443,
      "step": 9310
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.18623837043023e-05,
      "loss": 0.7558,
      "step": 9315
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.1778541366797888e-05,
      "loss": 0.7406,
      "step": 9320
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.1694840455924202e-05,
      "loss": 0.7999,
      "step": 9325
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.1611281123014858e-05,
      "loss": 0.7633,
      "step": 9330
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.1527863519147472e-05,
      "loss": 0.79,
      "step": 9335
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.1444587795143535e-05,
      "loss": 0.7535,
      "step": 9340
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.136145410156787e-05,
      "loss": 0.7685,
      "step": 9345
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.1278462588728575e-05,
      "loss": 0.7441,
      "step": 9350
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.1195613406676706e-05,
      "loss": 0.7542,
      "step": 9355
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.1112906705205936e-05,
      "loss": 0.755,
      "step": 9360
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.1030342633852307e-05,
      "loss": 0.7443,
      "step": 9365
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.0947921341894082e-05,
      "loss": 0.7795,
      "step": 9370
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.088208720964231e-05,
      "loss": 0.7806,
      "step": 9375
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.08163446277606e-05,
      "loss": 0.7737,
      "step": 9380
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.0734295258999025e-05,
      "loss": 0.7439,
      "step": 9385
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.065238920489496e-05,
      "loss": 0.7622,
      "step": 9390
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.0570626613536847e-05,
      "loss": 0.7953,
      "step": 9395
    },
    {
      "epoch": 2.41,
      "learning_rate": 2.0489007632753787e-05,
      "loss": 0.7815,
      "step": 9400
    },
    {
      "epoch": 2.41,
      "learning_rate": 2.040753241011525e-05,
      "loss": 0.7419,
      "step": 9405
    },
    {
      "epoch": 2.41,
      "learning_rate": 2.0326201092930764e-05,
      "loss": 0.7403,
      "step": 9410
    },
    {
      "epoch": 2.41,
      "learning_rate": 2.0245013828249625e-05,
      "loss": 0.7915,
      "step": 9415
    },
    {
      "epoch": 2.41,
      "learning_rate": 2.0163970762860783e-05,
      "loss": 0.7418,
      "step": 9420
    },
    {
      "epoch": 2.41,
      "learning_rate": 2.0083072043292407e-05,
      "loss": 0.7744,
      "step": 9425
    },
    {
      "epoch": 2.41,
      "learning_rate": 2.0002317815811645e-05,
      "loss": 0.7976,
      "step": 9430
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9921708226424495e-05,
      "loss": 0.7541,
      "step": 9435
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9841243420875398e-05,
      "loss": 0.7966,
      "step": 9440
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.976092354464699e-05,
      "loss": 0.7762,
      "step": 9445
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9680748742959887e-05,
      "loss": 0.7457,
      "step": 9450
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9600719160772473e-05,
      "loss": 0.7275,
      "step": 9455
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9520834942780495e-05,
      "loss": 0.7617,
      "step": 9460
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9457032327675385e-05,
      "loss": 0.7532,
      "step": 9465
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9377410129029406e-05,
      "loss": 0.7461,
      "step": 9470
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.9297933698327976e-05,
      "loss": 0.7516,
      "step": 9475
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.921860317926675e-05,
      "loss": 0.7863,
      "step": 9480
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.9139418715277546e-05,
      "loss": 0.7553,
      "step": 9485
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.9060380449528137e-05,
      "loss": 0.7239,
      "step": 9490
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.8981488524921996e-05,
      "loss": 0.7781,
      "step": 9495
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.8902743084097963e-05,
      "loss": 0.7677,
      "step": 9500
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.8824144269429998e-05,
      "loss": 0.7648,
      "step": 9505
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.874569222302708e-05,
      "loss": 0.7279,
      "step": 9510
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.8667387086732712e-05,
      "loss": 0.7506,
      "step": 9515
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.8589229002124786e-05,
      "loss": 0.7782,
      "step": 9520
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.8511218110515427e-05,
      "loss": 0.735,
      "step": 9525
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.8433354552950456e-05,
      "loss": 0.7552,
      "step": 9530
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.835563847020949e-05,
      "loss": 0.7474,
      "step": 9535
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.827807000280537e-05,
      "loss": 0.7508,
      "step": 9540
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.820064929098414e-05,
      "loss": 0.7394,
      "step": 9545
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.812337647472465e-05,
      "loss": 0.7584,
      "step": 9550
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.8046251693738327e-05,
      "loss": 0.7764,
      "step": 9555
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.796927508746903e-05,
      "loss": 0.7447,
      "step": 9560
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.789244679509262e-05,
      "loss": 0.7588,
      "step": 9565
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.7815766955516834e-05,
      "loss": 0.7661,
      "step": 9570
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.7739235707381064e-05,
      "loss": 0.7799,
      "step": 9575
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.766285318905595e-05,
      "loss": 0.7121,
      "step": 9580
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.7586619538643258e-05,
      "loss": 0.7543,
      "step": 9585
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.7510534893975673e-05,
      "loss": 0.7647,
      "step": 9590
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.7434599392616313e-05,
      "loss": 0.7636,
      "step": 9595
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.735881317185879e-05,
      "loss": 0.7493,
      "step": 9600
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.728317636872674e-05,
      "loss": 0.7475,
      "step": 9605
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.7207689119973646e-05,
      "loss": 0.7704,
      "step": 9610
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.71474070918498e-05,
      "loss": 0.7619,
      "step": 9615
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.7072189384734395e-05,
      "loss": 0.7394,
      "step": 9620
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.6997121613468458e-05,
      "loss": 0.7506,
      "step": 9625
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.6922203913776714e-05,
      "loss": 0.7359,
      "step": 9630
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.6847436421112393e-05,
      "loss": 0.7857,
      "step": 9635
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.677281927065735e-05,
      "loss": 0.7879,
      "step": 9640
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.6698352597321476e-05,
      "loss": 0.7077,
      "step": 9645
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.6624036535742627e-05,
      "loss": 0.7556,
      "step": 9650
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.6549871220286418e-05,
      "loss": 0.7445,
      "step": 9655
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.6475856785045794e-05,
      "loss": 0.748,
      "step": 9660
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.6401993363841038e-05,
      "loss": 0.727,
      "step": 9665
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.6328281090219277e-05,
      "loss": 0.7457,
      "step": 9670
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.62547200974544e-05,
      "loss": 0.7393,
      "step": 9675
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.6181310518546856e-05,
      "loss": 0.7839,
      "step": 9680
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.6108052486223147e-05,
      "loss": 0.7547,
      "step": 9685
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.6034946132935946e-05,
      "loss": 0.7612,
      "step": 9690
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.5961991590863678e-05,
      "loss": 0.732,
      "step": 9695
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.588918899191013e-05,
      "loss": 0.7343,
      "step": 9700
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.5816538467704567e-05,
      "loss": 0.7509,
      "step": 9705
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.574404014960117e-05,
      "loss": 0.8046,
      "step": 9710
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.5671694168678962e-05,
      "loss": 0.7523,
      "step": 9715
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.561392715462098e-05,
      "loss": 0.757,
      "step": 9720
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.5541855710064756e-05,
      "loss": 0.7218,
      "step": 9725
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.5469936968245013e-05,
      "loss": 0.7728,
      "step": 9730
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.5398171059192856e-05,
      "loss": 0.7402,
      "step": 9735
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.532655811266307e-05,
      "loss": 0.722,
      "step": 9740
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.5255098258133948e-05,
      "loss": 0.7417,
      "step": 9745
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.5198040687583803e-05,
      "loss": 0.7544,
      "step": 9750
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.512685672405737e-05,
      "loss": 0.7361,
      "step": 9755
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.5055826213597391e-05,
      "loss": 0.7386,
      "step": 9760
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.4984949284629046e-05,
      "loss": 0.7634,
      "step": 9765
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.4914226065299886e-05,
      "loss": 0.7345,
      "step": 9770
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.48436566834795e-05,
      "loss": 0.7916,
      "step": 9775
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.4773241266759296e-05,
      "loss": 0.7199,
      "step": 9780
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.4702979942452432e-05,
      "loss": 0.7367,
      "step": 9785
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.4632872837593348e-05,
      "loss": 0.7473,
      "step": 9790
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.4562920078937669e-05,
      "loss": 0.7712,
      "step": 9795
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.4493121792962027e-05,
      "loss": 0.7526,
      "step": 9800
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.4423478105863707e-05,
      "loss": 0.7523,
      "step": 9805
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.4353989143560475e-05,
      "loss": 0.7577,
      "step": 9810
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.428465503169033e-05,
      "loss": 0.7695,
      "step": 9815
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.4215475895611374e-05,
      "loss": 0.7761,
      "step": 9820
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.4146451860401445e-05,
      "loss": 0.7433,
      "step": 9825
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.4077583050857934e-05,
      "loss": 0.7675,
      "step": 9830
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.4008869591497665e-05,
      "loss": 0.7452,
      "step": 9835
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.3940311606556522e-05,
      "loss": 0.7225,
      "step": 9840
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.3871909219989266e-05,
      "loss": 0.7435,
      "step": 9845
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.3803662555469422e-05,
      "loss": 0.7199,
      "step": 9850
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.3749177426656478e-05,
      "loss": 0.7484,
      "step": 9855
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.3681211372580205e-05,
      "loss": 0.7479,
      "step": 9860
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.3613401385338464e-05,
      "loss": 0.7475,
      "step": 9865
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.354574758753363e-05,
      "loss": 0.727,
      "step": 9870
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.3478250101485746e-05,
      "loss": 0.7324,
      "step": 9875
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.3410909049232167e-05,
      "loss": 0.6984,
      "step": 9880
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.3343724552527414e-05,
      "loss": 0.7242,
      "step": 9885
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.327669673284303e-05,
      "loss": 0.7516,
      "step": 9890
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.3209825711367197e-05,
      "loss": 0.7372,
      "step": 9895
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.3156441870154191e-05,
      "loss": 0.7671,
      "step": 9900
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.3089853389942986e-05,
      "loss": 0.7611,
      "step": 9905
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.3023422045758415e-05,
      "loss": 0.7429,
      "step": 9910
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.2957147957710292e-05,
      "loss": 0.792,
      "step": 9915
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.2891031245624019e-05,
      "loss": 0.7879,
      "step": 9920
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.2825072029040496e-05,
      "loss": 0.7011,
      "step": 9925
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.2759270427215942e-05,
      "loss": 0.7574,
      "step": 9930
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.2693626559121418e-05,
      "loss": 0.74,
      "step": 9935
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.2628140543443001e-05,
      "loss": 0.7096,
      "step": 9940
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.2562812498581212e-05,
      "loss": 0.7617,
      "step": 9945
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.2510663881062656e-05,
      "loss": 0.7322,
      "step": 9950
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.2445620481128573e-05,
      "loss": 0.7158,
      "step": 9955
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.2380735382012576e-05,
      "loss": 0.761,
      "step": 9960
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.2316008701028847e-05,
      "loss": 0.7614,
      "step": 9965
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.2251440555205106e-05,
      "loss": 0.7601,
      "step": 9970
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.2187031061282327e-05,
      "loss": 0.7738,
      "step": 9975
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.2122780335714757e-05,
      "loss": 0.7127,
      "step": 9980
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.2058688494669546e-05,
      "loss": 0.7816,
      "step": 9985
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.1994755654026535e-05,
      "loss": 0.7538,
      "step": 9990
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.1930981929378171e-05,
      "loss": 0.7696,
      "step": 9995
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.1867367436029131e-05,
      "loss": 0.7597,
      "step": 10000
    },
    {
      "epoch": 2.56,
      "eval_loss": 1.5886551141738892,
      "eval_runtime": 2.0259,
      "eval_samples_per_second": 49.36,
      "eval_steps_per_second": 1.974,
      "step": 10000
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.1803912288996266e-05,
      "loss": 0.7156,
      "step": 10005
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.1740616603008293e-05,
      "loss": 0.7169,
      "step": 10010
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.1677480492505588e-05,
      "loss": 0.718,
      "step": 10015
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.1614504071640098e-05,
      "loss": 0.7268,
      "step": 10020
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.1551687454274906e-05,
      "loss": 0.7453,
      "step": 10025
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.148903075398431e-05,
      "loss": 0.8059,
      "step": 10030
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.1426534084053442e-05,
      "loss": 0.7176,
      "step": 10035
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.1364197557477974e-05,
      "loss": 0.7775,
      "step": 10040
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.1302021286964204e-05,
      "loss": 0.7312,
      "step": 10045
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.1240005384928576e-05,
      "loss": 0.7386,
      "step": 10050
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.1178149963497587e-05,
      "loss": 0.7418,
      "step": 10055
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.1116455134507664e-05,
      "loss": 0.7339,
      "step": 10060
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.105492100950477e-05,
      "loss": 0.7518,
      "step": 10065
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.0993547699744366e-05,
      "loss": 0.743,
      "step": 10070
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.0932335316191178e-05,
      "loss": 0.7235,
      "step": 10075
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.0871283969518908e-05,
      "loss": 0.7435,
      "step": 10080
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.0810393770110184e-05,
      "loss": 0.7282,
      "step": 10085
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.074966482805615e-05,
      "loss": 0.6981,
      "step": 10090
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.0689097253156532e-05,
      "loss": 0.754,
      "step": 10095
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.0628691154919213e-05,
      "loss": 0.7308,
      "step": 10100
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.0568446642560115e-05,
      "loss": 0.726,
      "step": 10105
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.0508363825003087e-05,
      "loss": 0.7712,
      "step": 10110
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.0448442810879555e-05,
      "loss": 0.7413,
      "step": 10115
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.0388683708528413e-05,
      "loss": 0.7368,
      "step": 10120
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.0329086625995843e-05,
      "loss": 0.7391,
      "step": 10125
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.0269651671035063e-05,
      "loss": 0.7216,
      "step": 10130
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.021037895110617e-05,
      "loss": 0.7444,
      "step": 10135
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.0151268573375916e-05,
      "loss": 0.7729,
      "step": 10140
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.009232064471759e-05,
      "loss": 0.7482,
      "step": 10145
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.0033535271710714e-05,
      "loss": 0.7665,
      "step": 10150
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.974912560640892e-06,
      "loss": 0.7547,
      "step": 10155
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.92813157961594e-06,
      "loss": 0.7654,
      "step": 10160
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.869801926933786e-06,
      "loss": 0.7369,
      "step": 10165
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.811635232223281e-06,
      "loss": 0.7335,
      "step": 10170
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.753631600651458e-06,
      "loss": 0.723,
      "step": 10175
    },
    {
      "epoch": 2.61,
      "learning_rate": 9.695791137090526e-06,
      "loss": 0.7534,
      "step": 10180
    },
    {
      "epoch": 2.61,
      "learning_rate": 9.638113946117722e-06,
      "loss": 0.7334,
      "step": 10185
    },
    {
      "epoch": 2.61,
      "learning_rate": 9.580600132015061e-06,
      "loss": 0.7401,
      "step": 10190
    },
    {
      "epoch": 2.61,
      "learning_rate": 9.534706781968339e-06,
      "loss": 0.7238,
      "step": 10195
    },
    {
      "epoch": 2.61,
      "learning_rate": 9.477487308079525e-06,
      "loss": 0.7591,
      "step": 10200
    },
    {
      "epoch": 2.61,
      "learning_rate": 9.420431501478411e-06,
      "loss": 0.7494,
      "step": 10205
    },
    {
      "epoch": 2.61,
      "learning_rate": 9.3635394653235e-06,
      "loss": 0.7431,
      "step": 10210
    },
    {
      "epoch": 2.61,
      "learning_rate": 9.318143820254332e-06,
      "loss": 0.7849,
      "step": 10215
    },
    {
      "epoch": 2.62,
      "learning_rate": 9.26154682991679e-06,
      "loss": 0.7534,
      "step": 10220
    },
    {
      "epoch": 2.62,
      "learning_rate": 9.20511389729345e-06,
      "loss": 0.7657,
      "step": 10225
    },
    {
      "epoch": 2.62,
      "learning_rate": 9.148845124416717e-06,
      "loss": 0.7238,
      "step": 10230
    },
    {
      "epoch": 2.62,
      "learning_rate": 9.092740613022099e-06,
      "loss": 0.7601,
      "step": 10235
    },
    {
      "epoch": 2.62,
      "learning_rate": 9.036800464548157e-06,
      "loss": 0.7155,
      "step": 10240
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.981024780136304e-06,
      "loss": 0.7555,
      "step": 10245
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.925413660630543e-06,
      "loss": 0.7489,
      "step": 10250
    },
    {
      "epoch": 2.63,
      "learning_rate": 8.869967206577378e-06,
      "loss": 0.7233,
      "step": 10255
    },
    {
      "epoch": 2.63,
      "learning_rate": 8.814685518225551e-06,
      "loss": 0.7043,
      "step": 10260
    },
    {
      "epoch": 2.63,
      "learning_rate": 8.759568695525988e-06,
      "loss": 0.7395,
      "step": 10265
    },
    {
      "epoch": 2.63,
      "learning_rate": 8.70461683813144e-06,
      "loss": 0.7285,
      "step": 10270
    },
    {
      "epoch": 2.63,
      "learning_rate": 8.649830045396435e-06,
      "loss": 0.7548,
      "step": 10275
    },
    {
      "epoch": 2.63,
      "learning_rate": 8.595208416377076e-06,
      "loss": 0.7466,
      "step": 10280
    },
    {
      "epoch": 2.63,
      "learning_rate": 8.540752049830825e-06,
      "loss": 0.7975,
      "step": 10285
    },
    {
      "epoch": 2.63,
      "learning_rate": 8.48646104421632e-06,
      "loss": 0.7699,
      "step": 10290
    },
    {
      "epoch": 2.64,
      "learning_rate": 8.4323354976933e-06,
      "loss": 0.7496,
      "step": 10295
    },
    {
      "epoch": 2.64,
      "learning_rate": 8.378375508122249e-06,
      "loss": 0.7521,
      "step": 10300
    },
    {
      "epoch": 2.64,
      "learning_rate": 8.324581173064394e-06,
      "loss": 0.7499,
      "step": 10305
    },
    {
      "epoch": 2.64,
      "learning_rate": 8.27095258978141e-06,
      "loss": 0.7918,
      "step": 10310
    },
    {
      "epoch": 2.64,
      "learning_rate": 8.217489855235339e-06,
      "loss": 0.7617,
      "step": 10315
    },
    {
      "epoch": 2.64,
      "learning_rate": 8.16419306608832e-06,
      "loss": 0.7163,
      "step": 10320
    },
    {
      "epoch": 2.64,
      "learning_rate": 8.111062318702433e-06,
      "loss": 0.7593,
      "step": 10325
    },
    {
      "epoch": 2.64,
      "learning_rate": 8.058097709139644e-06,
      "loss": 0.7391,
      "step": 10330
    },
    {
      "epoch": 2.65,
      "learning_rate": 8.005299333161453e-06,
      "loss": 0.7214,
      "step": 10335
    },
    {
      "epoch": 2.65,
      "learning_rate": 7.95266728622881e-06,
      "loss": 0.7695,
      "step": 10340
    },
    {
      "epoch": 2.65,
      "learning_rate": 7.900201663502016e-06,
      "loss": 0.7539,
      "step": 10345
    },
    {
      "epoch": 2.65,
      "learning_rate": 7.847902559840337e-06,
      "loss": 0.7683,
      "step": 10350
    },
    {
      "epoch": 2.65,
      "learning_rate": 7.795770069802088e-06,
      "loss": 0.7519,
      "step": 10355
    },
    {
      "epoch": 2.65,
      "learning_rate": 7.74380428764433e-06,
      "loss": 0.7332,
      "step": 10360
    },
    {
      "epoch": 2.65,
      "learning_rate": 7.692005307322626e-06,
      "loss": 0.7603,
      "step": 10365
    },
    {
      "epoch": 2.65,
      "learning_rate": 7.640373222491037e-06,
      "loss": 0.7235,
      "step": 10370
    },
    {
      "epoch": 2.66,
      "learning_rate": 7.588908126501859e-06,
      "loss": 0.7551,
      "step": 10375
    },
    {
      "epoch": 2.66,
      "learning_rate": 7.5376101124054136e-06,
      "loss": 0.702,
      "step": 10380
    },
    {
      "epoch": 2.66,
      "learning_rate": 7.48647927295002e-06,
      "loss": 0.7501,
      "step": 10385
    },
    {
      "epoch": 2.66,
      "learning_rate": 7.4355157005816634e-06,
      "loss": 0.7299,
      "step": 10390
    },
    {
      "epoch": 2.66,
      "learning_rate": 7.38471948744397e-06,
      "loss": 0.7368,
      "step": 10395
    },
    {
      "epoch": 2.66,
      "learning_rate": 7.334090725377918e-06,
      "loss": 0.75,
      "step": 10400
    },
    {
      "epoch": 2.66,
      "learning_rate": 7.283629505921752e-06,
      "loss": 0.7416,
      "step": 10405
    },
    {
      "epoch": 2.66,
      "learning_rate": 7.233335920310835e-06,
      "loss": 0.695,
      "step": 10410
    },
    {
      "epoch": 2.67,
      "learning_rate": 7.1832100594773525e-06,
      "loss": 0.6971,
      "step": 10415
    },
    {
      "epoch": 2.67,
      "learning_rate": 7.133252014050318e-06,
      "loss": 0.7392,
      "step": 10420
    },
    {
      "epoch": 2.67,
      "learning_rate": 7.083461874355335e-06,
      "loss": 0.7318,
      "step": 10425
    },
    {
      "epoch": 2.67,
      "learning_rate": 7.033839730414337e-06,
      "loss": 0.7247,
      "step": 10430
    },
    {
      "epoch": 2.67,
      "learning_rate": 6.984385671945626e-06,
      "loss": 0.701,
      "step": 10435
    },
    {
      "epoch": 2.67,
      "learning_rate": 6.935099788363531e-06,
      "loss": 0.7493,
      "step": 10440
    },
    {
      "epoch": 2.67,
      "learning_rate": 6.895792227308528e-06,
      "loss": 0.7523,
      "step": 10445
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.8565924007017e-06,
      "loss": 0.7656,
      "step": 10450
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.807744188101539e-06,
      "loss": 0.7349,
      "step": 10455
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.759064469760823e-06,
      "loss": 0.7774,
      "step": 10460
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.710553333693903e-06,
      "loss": 0.7456,
      "step": 10465
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.662210867610297e-06,
      "loss": 0.7106,
      "step": 10470
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.614037158914532e-06,
      "loss": 0.7537,
      "step": 10475
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.5660322947060925e-06,
      "loss": 0.7384,
      "step": 10480
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.5181963617791275e-06,
      "loss": 0.7542,
      "step": 10485
    },
    {
      "epoch": 2.69,
      "learning_rate": 6.470529446622353e-06,
      "loss": 0.7185,
      "step": 10490
    },
    {
      "epoch": 2.69,
      "learning_rate": 6.4230316354189195e-06,
      "loss": 0.7029,
      "step": 10495
    },
    {
      "epoch": 2.69,
      "learning_rate": 6.385155199022474e-06,
      "loss": 0.7254,
      "step": 10500
    },
    {
      "epoch": 2.69,
      "learning_rate": 6.337961991140551e-06,
      "loss": 0.7451,
      "step": 10505
    },
    {
      "epoch": 2.69,
      "learning_rate": 6.290938126897728e-06,
      "loss": 0.7559,
      "step": 10510
    },
    {
      "epoch": 2.69,
      "learning_rate": 6.244083691314484e-06,
      "loss": 0.7787,
      "step": 10515
    },
    {
      "epoch": 2.69,
      "learning_rate": 6.197398769104967e-06,
      "loss": 0.7704,
      "step": 10520
    },
    {
      "epoch": 2.69,
      "learning_rate": 6.1508834446768695e-06,
      "loss": 0.748,
      "step": 10525
    },
    {
      "epoch": 2.7,
      "learning_rate": 6.104537802131216e-06,
      "loss": 0.7112,
      "step": 10530
    },
    {
      "epoch": 2.7,
      "learning_rate": 6.058361925262235e-06,
      "loss": 0.7346,
      "step": 10535
    },
    {
      "epoch": 2.7,
      "learning_rate": 6.0123558975572645e-06,
      "loss": 0.7387,
      "step": 10540
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.9665198021965015e-06,
      "loss": 0.7576,
      "step": 10545
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.920853722052899e-06,
      "loss": 0.7444,
      "step": 10550
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.875357739692066e-06,
      "loss": 0.723,
      "step": 10555
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.830031937372004e-06,
      "loss": 0.7478,
      "step": 10560
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.784876397043071e-06,
      "loss": 0.7321,
      "step": 10565
    },
    {
      "epoch": 2.71,
      "learning_rate": 5.739891200347791e-06,
      "loss": 0.7317,
      "step": 10570
    },
    {
      "epoch": 2.71,
      "learning_rate": 5.69507642862066e-06,
      "loss": 0.7677,
      "step": 10575
    },
    {
      "epoch": 2.71,
      "learning_rate": 5.650432162888087e-06,
      "loss": 0.7331,
      "step": 10580
    },
    {
      "epoch": 2.71,
      "learning_rate": 5.605958483868135e-06,
      "loss": 0.7325,
      "step": 10585
    },
    {
      "epoch": 2.71,
      "learning_rate": 5.561655471970539e-06,
      "loss": 0.7291,
      "step": 10590
    },
    {
      "epoch": 2.71,
      "learning_rate": 5.517523207296382e-06,
      "loss": 0.7529,
      "step": 10595
    },
    {
      "epoch": 2.71,
      "learning_rate": 5.473561769638058e-06,
      "loss": 0.7533,
      "step": 10600
    },
    {
      "epoch": 2.71,
      "learning_rate": 5.429771238479109e-06,
      "loss": 0.7306,
      "step": 10605
    },
    {
      "epoch": 2.72,
      "learning_rate": 5.3861516929940794e-06,
      "loss": 0.7554,
      "step": 10610
    },
    {
      "epoch": 2.72,
      "learning_rate": 5.35137921929918e-06,
      "loss": 0.764,
      "step": 10615
    },
    {
      "epoch": 2.72,
      "learning_rate": 5.3080676465600885e-06,
      "loss": 0.7494,
      "step": 10620
    },
    {
      "epoch": 2.72,
      "learning_rate": 5.264927279538467e-06,
      "loss": 0.7426,
      "step": 10625
    },
    {
      "epoch": 2.72,
      "learning_rate": 5.221958196233345e-06,
      "loss": 0.749,
      "step": 10630
    },
    {
      "epoch": 2.72,
      "learning_rate": 5.179160474334043e-06,
      "loss": 0.7352,
      "step": 10635
    },
    {
      "epoch": 2.72,
      "learning_rate": 5.136534191220044e-06,
      "loss": 0.7363,
      "step": 10640
    },
    {
      "epoch": 2.72,
      "learning_rate": 5.094079423960918e-06,
      "loss": 0.7293,
      "step": 10645
    },
    {
      "epoch": 2.73,
      "learning_rate": 5.051796249316076e-06,
      "loss": 0.7097,
      "step": 10650
    },
    {
      "epoch": 2.73,
      "learning_rate": 5.00968474373471e-06,
      "loss": 0.7339,
      "step": 10655
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.967744983355638e-06,
      "loss": 0.7604,
      "step": 10660
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.925977044007113e-06,
      "loss": 0.7612,
      "step": 10665
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.884381001206795e-06,
      "loss": 0.7062,
      "step": 10670
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.8429569301614795e-06,
      "loss": 0.759,
      "step": 10675
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.8017049057671124e-06,
      "loss": 0.735,
      "step": 10680
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.76062500260851e-06,
      "loss": 0.7663,
      "step": 10685
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.719717294959303e-06,
      "loss": 0.7351,
      "step": 10690
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.678981856781805e-06,
      "loss": 0.7365,
      "step": 10695
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.638418761726859e-06,
      "loss": 0.748,
      "step": 10700
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.598028083133676e-06,
      "loss": 0.7681,
      "step": 10705
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.557809894029807e-06,
      "loss": 0.7592,
      "step": 10710
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.517764267130808e-06,
      "loss": 0.7178,
      "step": 10715
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.477891274840362e-06,
      "loss": 0.7166,
      "step": 10720
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.438190989250002e-06,
      "loss": 0.7478,
      "step": 10725
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.40655515784788e-06,
      "loss": 0.7539,
      "step": 10730
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.367165924991401e-06,
      "loss": 0.7454,
      "step": 10735
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.327949599029568e-06,
      "loss": 0.7385,
      "step": 10740
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.288906250866609e-06,
      "loss": 0.7159,
      "step": 10745
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.250035951094056e-06,
      "loss": 0.7283,
      "step": 10750
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.21133876999047e-06,
      "loss": 0.7306,
      "step": 10755
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.172814777521483e-06,
      "loss": 0.7292,
      "step": 10760
    },
    {
      "epoch": 2.76,
      "learning_rate": 4.134464043339592e-06,
      "loss": 0.7363,
      "step": 10765
    },
    {
      "epoch": 2.76,
      "learning_rate": 4.103908248567223e-06,
      "loss": 0.7536,
      "step": 10770
    },
    {
      "epoch": 2.76,
      "learning_rate": 4.065869553826307e-06,
      "loss": 0.7517,
      "step": 10775
    },
    {
      "epoch": 2.76,
      "learning_rate": 4.028004310732491e-06,
      "loss": 0.7423,
      "step": 10780
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.99031258774718e-06,
      "loss": 0.7507,
      "step": 10785
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.952794453018149e-06,
      "loss": 0.7566,
      "step": 10790
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.915449974379204e-06,
      "loss": 0.7348,
      "step": 10795
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.878279219350245e-06,
      "loss": 0.7017,
      "step": 10800
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.84128225513708e-06,
      "loss": 0.7381,
      "step": 10805
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.804459148631234e-06,
      "loss": 0.7061,
      "step": 10810
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.767809966409963e-06,
      "loss": 0.7511,
      "step": 10815
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.7313347747360393e-06,
      "loss": 0.7428,
      "step": 10820
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.6950336395576322e-06,
      "loss": 0.7484,
      "step": 10825
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.6589066265082404e-06,
      "loss": 0.7195,
      "step": 10830
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.6229538009065476e-06,
      "loss": 0.7526,
      "step": 10835
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.5871752277562566e-06,
      "loss": 0.7134,
      "step": 10840
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.5515709717460876e-06,
      "loss": 0.6897,
      "step": 10845
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.5161410972495147e-06,
      "loss": 0.7358,
      "step": 10850
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.480885668324807e-06,
      "loss": 0.7443,
      "step": 10855
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.4458047487147315e-06,
      "loss": 0.7133,
      "step": 10860
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.4108984018466406e-06,
      "loss": 0.742,
      "step": 10865
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.3761666908321942e-06,
      "loss": 0.7426,
      "step": 10870
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.3416096784672836e-06,
      "loss": 0.7388,
      "step": 10875
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.3140898936002806e-06,
      "loss": 0.763,
      "step": 10880
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.286681987706497e-06,
      "loss": 0.7064,
      "step": 10885
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.2525794834484593e-06,
      "loss": 0.7621,
      "step": 10890
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.2186519012891827e-06,
      "loss": 0.7613,
      "step": 10895
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.1848993025707206e-06,
      "loss": 0.7191,
      "step": 10900
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.151321748318692e-06,
      "loss": 0.7314,
      "step": 10905
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.1179192992423332e-06,
      "loss": 0.691,
      "step": 10910
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.091323456302142e-06,
      "loss": 0.7511,
      "step": 10915
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.058236348518406e-06,
      "loss": 0.7268,
      "step": 10920
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.0253245142113807e-06,
      "loss": 0.7421,
      "step": 10925
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.9925880128865657e-06,
      "loss": 0.7441,
      "step": 10930
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.9600269037325377e-06,
      "loss": 0.7378,
      "step": 10935
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.9276412456206826e-06,
      "loss": 0.7531,
      "step": 10940
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.8954310971051946e-06,
      "loss": 0.7232,
      "step": 10945
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.8633965164229672e-06,
      "loss": 0.7461,
      "step": 10950
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.8315375614934137e-06,
      "loss": 0.6969,
      "step": 10955
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.7998542899184355e-06,
      "loss": 0.7016,
      "step": 10960
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.7683467589822874e-06,
      "loss": 0.7524,
      "step": 10965
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.7370150256515127e-06,
      "loss": 0.7248,
      "step": 10970
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.7058591465747517e-06,
      "loss": 0.7473,
      "step": 10975
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.6748791780827344e-06,
      "loss": 0.7665,
      "step": 10980
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.6440751761881323e-06,
      "loss": 0.713,
      "step": 10985
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.61344719658545e-06,
      "loss": 0.7783,
      "step": 10990
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.5829952946509585e-06,
      "loss": 0.7176,
      "step": 10995
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.5527195254425596e-06,
      "loss": 0.715,
      "step": 11000
    },
    {
      "epoch": 2.82,
      "eval_loss": 1.5941790342330933,
      "eval_runtime": 2.0287,
      "eval_samples_per_second": 49.291,
      "eval_steps_per_second": 1.972,
      "step": 11000
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.5226199436997113e-06,
      "loss": 0.7453,
      "step": 11005
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.4926966038433254e-06,
      "loss": 0.7323,
      "step": 11010
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.462949559975625e-06,
      "loss": 0.734,
      "step": 11015
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.433378865880154e-06,
      "loss": 0.7219,
      "step": 11020
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.403984575021556e-06,
      "loss": 0.7302,
      "step": 11025
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.374766740545531e-06,
      "loss": 0.7524,
      "step": 11030
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.3457254152787766e-06,
      "loss": 0.7725,
      "step": 11035
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.316860651728836e-06,
      "loss": 0.7354,
      "step": 11040
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.2881725020840184e-06,
      "loss": 0.7248,
      "step": 11045
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.259661018213333e-06,
      "loss": 0.7433,
      "step": 11050
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.2313262516663436e-06,
      "loss": 0.7765,
      "step": 11055
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.2031682536731158e-06,
      "loss": 0.7678,
      "step": 11060
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.175187075144136e-06,
      "loss": 0.6898,
      "step": 11065
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.1473827666701364e-06,
      "loss": 0.7526,
      "step": 11070
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.119755378522137e-06,
      "loss": 0.7479,
      "step": 11075
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.0923049606512147e-06,
      "loss": 0.7417,
      "step": 11080
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.0650315626885243e-06,
      "loss": 0.733,
      "step": 11085
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.037935233945154e-06,
      "loss": 0.7329,
      "step": 11090
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.0110160234120156e-06,
      "loss": 0.752,
      "step": 11095
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.984273979759843e-06,
      "loss": 0.754,
      "step": 11100
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.9577091513389933e-06,
      "loss": 0.7595,
      "step": 11105
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.9313215861794352e-06,
      "loss": 0.7345,
      "step": 11110
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.905111331990661e-06,
      "loss": 0.7556,
      "step": 11115
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.8790784361615189e-06,
      "loss": 0.7045,
      "step": 11120
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.8583798491577876e-06,
      "loss": 0.7597,
      "step": 11125
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.8326663167724024e-06,
      "loss": 0.7237,
      "step": 11130
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.8071302737293295e-06,
      "loss": 0.7016,
      "step": 11135
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.786829262658285e-06,
      "loss": 0.7563,
      "step": 11140
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.7616128165638179e-06,
      "loss": 0.7167,
      "step": 11145
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.7365739882784116e-06,
      "loss": 0.7622,
      "step": 11150
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.7117128230729862e-06,
      "loss": 0.752,
      "step": 11155
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.6870293658972525e-06,
      "loss": 0.7209,
      "step": 11160
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.662523661379578e-06,
      "loss": 0.705,
      "step": 11165
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.63819575382701e-06,
      "loss": 0.7215,
      "step": 11170
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.6140456872250742e-06,
      "loss": 0.747,
      "step": 11175
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.5900735052377992e-06,
      "loss": 0.7034,
      "step": 11180
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.5662792512075586e-06,
      "loss": 0.7534,
      "step": 11185
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.5426629681550509e-06,
      "loss": 0.7393,
      "step": 11190
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.5192246987791981e-06,
      "loss": 0.6835,
      "step": 11195
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.4959644854570465e-06,
      "loss": 0.7143,
      "step": 11200
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.4728823702437334e-06,
      "loss": 0.7465,
      "step": 11205
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.4499783948723756e-06,
      "loss": 0.6884,
      "step": 11210
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.4272526007540144e-06,
      "loss": 0.7223,
      "step": 11215
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.4047050289775598e-06,
      "loss": 0.7425,
      "step": 11220
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.3823357203096243e-06,
      "loss": 0.7353,
      "step": 11225
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.3601447151945889e-06,
      "loss": 0.7128,
      "step": 11230
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.338132053754404e-06,
      "loss": 0.7654,
      "step": 11235
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.3162977757886108e-06,
      "loss": 0.7225,
      "step": 11240
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.2946419207741867e-06,
      "loss": 0.7191,
      "step": 11245
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.2774457276102381e-06,
      "loss": 0.7095,
      "step": 11250
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.2561111323605712e-06,
      "loss": 0.7442,
      "step": 11255
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.2391719971949435e-06,
      "loss": 0.7712,
      "step": 11260
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.218158786688306e-06,
      "loss": 0.7304,
      "step": 11265
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.1973241765711352e-06,
      "loss": 0.7266,
      "step": 11270
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.1766682045130207e-06,
      "loss": 0.788,
      "step": 11275
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.1561909078605215e-06,
      "loss": 0.7357,
      "step": 11280
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.1358923236371888e-06,
      "loss": 0.7393,
      "step": 11285
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.115772488543454e-06,
      "loss": 0.7702,
      "step": 11290
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.0958314389565294e-06,
      "loss": 0.7311,
      "step": 11295
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.0760692109304304e-06,
      "loss": 0.7572,
      "step": 11300
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.0564858401957979e-06,
      "loss": 0.7633,
      "step": 11305
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.0370813621599528e-06,
      "loss": 0.7534,
      "step": 11310
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.0178558119067315e-06,
      "loss": 0.7509,
      "step": 11315
    },
    {
      "epoch": 2.9,
      "learning_rate": 9.98809224196473e-07,
      "loss": 0.7378,
      "step": 11320
    },
    {
      "epoch": 2.9,
      "learning_rate": 9.79941633465964e-07,
      "loss": 0.7193,
      "step": 11325
    },
    {
      "epoch": 2.9,
      "learning_rate": 9.612530738283387e-07,
      "loss": 0.7503,
      "step": 11330
    },
    {
      "epoch": 2.9,
      "learning_rate": 9.427435790730576e-07,
      "loss": 0.7165,
      "step": 11335
    },
    {
      "epoch": 2.9,
      "learning_rate": 9.244131826658065e-07,
      "loss": 0.7455,
      "step": 11340
    },
    {
      "epoch": 2.9,
      "learning_rate": 9.062619177484632e-07,
      "loss": 0.7504,
      "step": 11345
    },
    {
      "epoch": 2.91,
      "learning_rate": 8.88289817139043e-07,
      "loss": 0.7197,
      "step": 11350
    },
    {
      "epoch": 2.91,
      "learning_rate": 8.704969133316087e-07,
      "loss": 0.7668,
      "step": 11355
    },
    {
      "epoch": 2.91,
      "learning_rate": 8.52883238496227e-07,
      "loss": 0.7303,
      "step": 11360
    },
    {
      "epoch": 2.91,
      "learning_rate": 8.354488244789572e-07,
      "loss": 0.721,
      "step": 11365
    },
    {
      "epoch": 2.91,
      "learning_rate": 8.181937028016951e-07,
      "loss": 0.7242,
      "step": 11370
    },
    {
      "epoch": 2.91,
      "learning_rate": 8.04518716920466e-07,
      "loss": 0.7498,
      "step": 11375
    },
    {
      "epoch": 2.91,
      "learning_rate": 7.909585139335041e-07,
      "loss": 0.7349,
      "step": 11380
    },
    {
      "epoch": 2.91,
      "learning_rate": 7.741696975364643e-07,
      "loss": 0.7437,
      "step": 11385
    },
    {
      "epoch": 2.92,
      "learning_rate": 7.575602842738616e-07,
      "loss": 0.7473,
      "step": 11390
    },
    {
      "epoch": 2.92,
      "learning_rate": 7.411303041759632e-07,
      "loss": 0.7231,
      "step": 11395
    },
    {
      "epoch": 2.92,
      "learning_rate": 7.248797869486623e-07,
      "loss": 0.7626,
      "step": 11400
    },
    {
      "epoch": 2.92,
      "learning_rate": 7.088087619733453e-07,
      "loss": 0.7439,
      "step": 11405
    },
    {
      "epoch": 2.92,
      "learning_rate": 6.92917258306891e-07,
      "loss": 0.7556,
      "step": 11410
    },
    {
      "epoch": 2.92,
      "learning_rate": 6.77205304681594e-07,
      "loss": 0.7185,
      "step": 11415
    },
    {
      "epoch": 2.92,
      "learning_rate": 6.616729295051083e-07,
      "loss": 0.7589,
      "step": 11420
    },
    {
      "epoch": 2.92,
      "learning_rate": 6.463201608604031e-07,
      "loss": 0.7593,
      "step": 11425
    },
    {
      "epoch": 2.93,
      "learning_rate": 6.311470265057517e-07,
      "loss": 0.7631,
      "step": 11430
    },
    {
      "epoch": 2.93,
      "learning_rate": 6.161535538745878e-07,
      "loss": 0.7323,
      "step": 11435
    },
    {
      "epoch": 2.93,
      "learning_rate": 6.013397700755375e-07,
      "loss": 0.7328,
      "step": 11440
    },
    {
      "epoch": 2.93,
      "learning_rate": 5.867057018923761e-07,
      "loss": 0.7285,
      "step": 11445
    },
    {
      "epoch": 2.93,
      "learning_rate": 5.722513757838722e-07,
      "loss": 0.7268,
      "step": 11450
    },
    {
      "epoch": 2.93,
      "learning_rate": 5.579768178838984e-07,
      "loss": 0.7321,
      "step": 11455
    },
    {
      "epoch": 2.93,
      "learning_rate": 5.438820540012435e-07,
      "loss": 0.7657,
      "step": 11460
    },
    {
      "epoch": 2.93,
      "learning_rate": 5.299671096196334e-07,
      "loss": 0.7271,
      "step": 11465
    },
    {
      "epoch": 2.94,
      "learning_rate": 5.162320098977103e-07,
      "loss": 0.7236,
      "step": 11470
    },
    {
      "epoch": 2.94,
      "learning_rate": 5.026767796689091e-07,
      "loss": 0.7433,
      "step": 11475
    },
    {
      "epoch": 2.94,
      "learning_rate": 4.89301443441481e-07,
      "loss": 0.7542,
      "step": 11480
    },
    {
      "epoch": 2.94,
      "learning_rate": 4.761060253984151e-07,
      "loss": 0.7277,
      "step": 11485
    },
    {
      "epoch": 2.94,
      "learning_rate": 4.630905493973936e-07,
      "loss": 0.7282,
      "step": 11490
    },
    {
      "epoch": 2.94,
      "learning_rate": 4.502550389707927e-07,
      "loss": 0.736,
      "step": 11495
    },
    {
      "epoch": 2.94,
      "learning_rate": 4.375995173255598e-07,
      "loss": 0.732,
      "step": 11500
    },
    {
      "epoch": 2.95,
      "learning_rate": 4.251240073432361e-07,
      "loss": 0.7365,
      "step": 11505
    },
    {
      "epoch": 2.95,
      "learning_rate": 4.1282853157992295e-07,
      "loss": 0.7325,
      "step": 11510
    },
    {
      "epoch": 2.95,
      "learning_rate": 4.0071311226617116e-07,
      "loss": 0.768,
      "step": 11515
    },
    {
      "epoch": 2.95,
      "learning_rate": 3.88777771307014e-07,
      "loss": 0.7324,
      "step": 11520
    },
    {
      "epoch": 2.95,
      "learning_rate": 3.793591694673082e-07,
      "loss": 0.7499,
      "step": 11525
    },
    {
      "epoch": 2.95,
      "learning_rate": 3.677480237077813e-07,
      "loss": 0.7301,
      "step": 11530
    },
    {
      "epoch": 2.95,
      "learning_rate": 3.5631701590467246e-07,
      "loss": 0.7656,
      "step": 11535
    },
    {
      "epoch": 2.95,
      "learning_rate": 3.450661667255495e-07,
      "loss": 0.7299,
      "step": 11540
    },
    {
      "epoch": 2.96,
      "learning_rate": 3.3399549651229603e-07,
      "loss": 0.7466,
      "step": 11545
    },
    {
      "epoch": 2.96,
      "learning_rate": 3.231050252809897e-07,
      "loss": 0.7071,
      "step": 11550
    },
    {
      "epoch": 2.96,
      "learning_rate": 3.145224048057727e-07,
      "loss": 0.7183,
      "step": 11555
    },
    {
      "epoch": 2.96,
      "learning_rate": 3.0395634114250836e-07,
      "loss": 0.7307,
      "step": 11560
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.9357053077285755e-07,
      "loss": 0.764,
      "step": 11565
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.8336499247463287e-07,
      "loss": 0.7169,
      "step": 11570
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.733397446997632e-07,
      "loss": 0.6721,
      "step": 11575
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.634948055741715e-07,
      "loss": 0.7416,
      "step": 11580
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.53830192897786e-07,
      "loss": 0.704,
      "step": 11585
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.4434592414450676e-07,
      "loss": 0.711,
      "step": 11590
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.3504201646215028e-07,
      "loss": 0.7354,
      "step": 11595
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.2591848667247173e-07,
      "loss": 0.7408,
      "step": 11600
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.1697535127105372e-07,
      "loss": 0.7419,
      "step": 11605
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.0821262642733985e-07,
      "loss": 0.7498,
      "step": 11610
    },
    {
      "epoch": 2.97,
      "learning_rate": 1.9963032798460125e-07,
      "loss": 0.7346,
      "step": 11615
    },
    {
      "epoch": 2.97,
      "learning_rate": 1.912284714598811e-07,
      "loss": 0.7794,
      "step": 11620
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.8300707204395028e-07,
      "loss": 0.7755,
      "step": 11625
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.7496614460135176e-07,
      "loss": 0.7596,
      "step": 11630
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.6710570367031164e-07,
      "loss": 0.7305,
      "step": 11635
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.594257634627283e-07,
      "loss": 0.7371,
      "step": 11640
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.5192633786415e-07,
      "loss": 0.7111,
      "step": 11645
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.4460744043376385e-07,
      "loss": 0.7592,
      "step": 11650
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.3746908440432915e-07,
      "loss": 0.7156,
      "step": 11655
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.318883980735097e-07,
      "loss": 0.7421,
      "step": 11660
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.2507504887044884e-07,
      "loss": 0.7729,
      "step": 11665
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.197543841632398e-07,
      "loss": 0.7528,
      "step": 11670
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.1326608169920372e-07,
      "loss": 0.7502,
      "step": 11675
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.0695837730220203e-07,
      "loss": 0.7245,
      "step": 11680
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.0083128237673434e-07,
      "loss": 0.7325,
      "step": 11685
    },
    {
      "epoch": 2.99,
      "learning_rate": 9.488480800077249e-08,
      "loss": 0.7045,
      "step": 11690
    },
    {
      "epoch": 2.99,
      "learning_rate": 8.911896492571625e-08,
      "loss": 0.7447,
      "step": 11695
    },
    {
      "epoch": 3.0,
      "learning_rate": 8.353376357636001e-08,
      "loss": 0.7317,
      "step": 11700
    },
    {
      "epoch": 3.0,
      "learning_rate": 7.8129214050926e-08,
      "loss": 0.7322,
      "step": 11705
    },
    {
      "epoch": 3.0,
      "learning_rate": 7.290532612096445e-08,
      "loss": 0.7227,
      "step": 11710
    },
    {
      "epoch": 3.0,
      "learning_rate": 6.786210923145353e-08,
      "loss": 0.7408,
      "step": 11715
    },
    {
      "epoch": 3.0,
      "step": 11718,
      "total_flos": 995899367424000.0,
      "train_loss": 1.3950741625133278,
      "train_runtime": 240198.6559,
      "train_samples_per_second": 6.245,
      "train_steps_per_second": 0.049
    }
  ],
  "max_steps": 11718,
  "num_train_epochs": 3,
  "total_flos": 995899367424000.0,
  "trial_name": null,
  "trial_params": null
}