{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 2000,
  "global_step": 21811,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 2e-05,
      "loss": 5.9866,
      "step": 10
    },
    {
      "epoch": 0.0,
      "learning_rate": 4e-05,
      "loss": 5.7802,
      "step": 20
    },
    {
      "epoch": 0.0,
      "learning_rate": 6e-05,
      "loss": 5.2475,
      "step": 30
    },
    {
      "epoch": 0.0,
      "learning_rate": 8e-05,
      "loss": 4.6122,
      "step": 40
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0001,
      "loss": 4.2112,
      "step": 50
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00012,
      "loss": 3.7987,
      "step": 60
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00014,
      "loss": 3.5017,
      "step": 70
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00016,
      "loss": 3.3313,
      "step": 80
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00018,
      "loss": 3.2171,
      "step": 90
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0002,
      "loss": 3.1014,
      "step": 100
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00019999989530883824,
      "loss": 3.0089,
      "step": 110
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00019999958123557212,
      "loss": 2.9289,
      "step": 120
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001999990577808593,
      "loss": 2.9056,
      "step": 130
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00019999832494579575,
      "loss": 2.8754,
      "step": 140
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00019999738273191592,
      "loss": 2.838,
      "step": 150
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001999962311411927,
      "loss": 2.8294,
      "step": 160
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001999948701760372,
      "loss": 2.7845,
      "step": 170
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00019999329983929912,
      "loss": 2.7783,
      "step": 180
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00019999152013426644,
      "loss": 2.7443,
      "step": 190
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00019998953106466558,
      "loss": 2.7478,
      "step": 200
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00019998733263466124,
      "loss": 2.7408,
      "step": 210
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00019998492484885663,
      "loss": 2.7232,
      "step": 220
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00019998230771229314,
      "loss": 2.6969,
      "step": 230
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00019997948123045065,
      "loss": 2.6979,
      "step": 240
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00019997644540924728,
      "loss": 2.6859,
      "step": 250
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001999732002550395,
      "loss": 2.6947,
      "step": 260
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00019996974577462213,
      "loss": 2.6549,
      "step": 270
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00019996608197522822,
      "loss": 2.6624,
      "step": 280
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00019996220886452911,
      "loss": 2.6462,
      "step": 290
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001999581264506344,
      "loss": 2.6612,
      "step": 300
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.000199953834742092,
      "loss": 2.6211,
      "step": 310
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00019994933374788788,
      "loss": 2.642,
      "step": 320
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019994462347744645,
      "loss": 2.6373,
      "step": 330
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001999397039406301,
      "loss": 2.6165,
      "step": 340
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001999345751477395,
      "loss": 2.6179,
      "step": 350
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019992923710951342,
      "loss": 2.6159,
      "step": 360
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001999236898371288,
      "loss": 2.606,
      "step": 370
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019991793334220065,
      "loss": 2.604,
      "step": 380
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.000199911967636782,
      "loss": 2.6028,
      "step": 390
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019990579273336398,
      "loss": 2.6024,
      "step": 400
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019989940864487585,
      "loss": 2.5923,
      "step": 410
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019989281538468465,
      "loss": 2.6114,
      "step": 420
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019988601296659555,
      "loss": 2.5789,
      "step": 430
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019987900140485163,
      "loss": 2.6063,
      "step": 440
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001998717807141338,
      "loss": 2.5787,
      "step": 450
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019986435090956097,
      "loss": 2.555,
      "step": 460
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019985671200668982,
      "loss": 2.5754,
      "step": 470
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019984886402151482,
      "loss": 2.5737,
      "step": 480
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019984080697046835,
      "loss": 2.5746,
      "step": 490
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019983254087042036,
      "loss": 2.5604,
      "step": 500
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019982406573867864,
      "loss": 2.564,
      "step": 510
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019981538159298868,
      "loss": 2.5971,
      "step": 520
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019980648845153342,
      "loss": 2.5353,
      "step": 530
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019979738633293362,
      "loss": 2.5716,
      "step": 540
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0001997880752562475,
      "loss": 2.5572,
      "step": 550
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019977855524097075,
      "loss": 2.5381,
      "step": 560
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019976882630703667,
      "loss": 2.5477,
      "step": 570
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019975888847481588,
      "loss": 2.5408,
      "step": 580
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019974874176511648,
      "loss": 2.5406,
      "step": 590
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019973838619918383,
      "loss": 2.569,
      "step": 600
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0001997278217987007,
      "loss": 2.5659,
      "step": 610
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0001997170485857871,
      "loss": 2.5437,
      "step": 620
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019970606658300017,
      "loss": 2.549,
      "step": 630
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019969487581333436,
      "loss": 2.5364,
      "step": 640
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019968347630022106,
      "loss": 2.5291,
      "step": 650
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019967186806752892,
      "loss": 2.5358,
      "step": 660
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019966005113956353,
      "loss": 2.5367,
      "step": 670
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019964802554106736,
      "loss": 2.5288,
      "step": 680
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019963579129721998,
      "loss": 2.5281,
      "step": 690
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0001996233484336377,
      "loss": 2.5068,
      "step": 700
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019961069697637366,
      "loss": 2.5263,
      "step": 710
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019959783695191783,
      "loss": 2.5378,
      "step": 720
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019958476838719674,
      "loss": 2.5527,
      "step": 730
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019957149130957378,
      "loss": 2.5342,
      "step": 740
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019955800574684868,
      "loss": 2.5549,
      "step": 750
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019954431172725787,
      "loss": 2.4973,
      "step": 760
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019953040927947423,
      "loss": 2.5182,
      "step": 770
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019951629843260697,
      "loss": 2.53,
      "step": 780
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001995019792162018,
      "loss": 2.5162,
      "step": 790
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019948745166024055,
      "loss": 2.4986,
      "step": 800
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019947271579514138,
      "loss": 2.5029,
      "step": 810
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019945777165175858,
      "loss": 2.5156,
      "step": 820
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019944261926138258,
      "loss": 2.5139,
      "step": 830
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019942725865573974,
      "loss": 2.5217,
      "step": 840
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001994116898669925,
      "loss": 2.5289,
      "step": 850
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019939591292773913,
      "loss": 2.4894,
      "step": 860
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019937992787101382,
      "loss": 2.5175,
      "step": 870
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019936373473028636,
      "loss": 2.4996,
      "step": 880
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019934733353946235,
      "loss": 2.5087,
      "step": 890
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.000199330724332883,
      "loss": 2.4815,
      "step": 900
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019931390714532507,
      "loss": 2.4905,
      "step": 910
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019929688201200075,
      "loss": 2.511,
      "step": 920
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019927964896855767,
      "loss": 2.5164,
      "step": 930
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019926220805107877,
      "loss": 2.4972,
      "step": 940
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019924455929608224,
      "loss": 2.4919,
      "step": 950
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019922670274052147,
      "loss": 2.5115,
      "step": 960
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019920863842178493,
      "loss": 2.5169,
      "step": 970
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019919036637769613,
      "loss": 2.4865,
      "step": 980
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019917188664651345,
      "loss": 2.5039,
      "step": 990
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019915319926693022,
      "loss": 2.518,
      "step": 1000
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019913430427807448,
      "loss": 2.4838,
      "step": 1010
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019911520171950902,
      "loss": 2.5011,
      "step": 1020
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019909589163123124,
      "loss": 2.4865,
      "step": 1030
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019907637405367298,
      "loss": 2.4865,
      "step": 1040
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0001990566490277007,
      "loss": 2.4835,
      "step": 1050
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019903671659461504,
      "loss": 2.4845,
      "step": 1060
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019901657679615098,
      "loss": 2.4848,
      "step": 1070
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019899622967447776,
      "loss": 2.4871,
      "step": 1080
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019897567527219862,
      "loss": 2.4848,
      "step": 1090
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019895491363235084,
      "loss": 2.4875,
      "step": 1100
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019893394479840565,
      "loss": 2.503,
      "step": 1110
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0001989127688142681,
      "loss": 2.4778,
      "step": 1120
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019889138572427685,
      "loss": 2.4976,
      "step": 1130
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019886979557320444,
      "loss": 2.4647,
      "step": 1140
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019884799840625677,
      "loss": 2.4749,
      "step": 1150
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019882599426907326,
      "loss": 2.4967,
      "step": 1160
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0001988037832077267,
      "loss": 2.4628,
      "step": 1170
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0001987813652687231,
      "loss": 2.4912,
      "step": 1180
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019875874049900166,
      "loss": 2.4981,
      "step": 1190
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019873590894593465,
      "loss": 2.4628,
      "step": 1200
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019871287065732735,
      "loss": 2.493,
      "step": 1210
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019868962568141783,
      "loss": 2.4585,
      "step": 1220
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019866617406687692,
      "loss": 2.4467,
      "step": 1230
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019864251586280824,
      "loss": 2.4778,
      "step": 1240
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019861865111874783,
      "loss": 2.485,
      "step": 1250
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019859457988466426,
      "loss": 2.4815,
      "step": 1260
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019857030221095845,
      "loss": 2.4662,
      "step": 1270
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0001985458181484635,
      "loss": 2.4743,
      "step": 1280
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0001985211277484448,
      "loss": 2.471,
      "step": 1290
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019849623106259966,
      "loss": 2.4714,
      "step": 1300
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019847112814305726,
      "loss": 2.471,
      "step": 1310
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0001984458190423788,
      "loss": 2.4944,
      "step": 1320
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019842030381355693,
      "loss": 2.4824,
      "step": 1330
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019839458251001615,
      "loss": 2.4546,
      "step": 1340
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019836865518561223,
      "loss": 2.4587,
      "step": 1350
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019834252189463247,
      "loss": 2.4773,
      "step": 1360
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019831618269179532,
      "loss": 2.4772,
      "step": 1370
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019828963763225045,
      "loss": 2.4833,
      "step": 1380
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019826288677157847,
      "loss": 2.4767,
      "step": 1390
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019823593016579102,
      "loss": 2.4637,
      "step": 1400
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019820876787133041,
      "loss": 2.4569,
      "step": 1410
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019818139994506976,
      "loss": 2.467,
      "step": 1420
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019815382644431257,
      "loss": 2.4592,
      "step": 1430
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019812604742679293,
      "loss": 2.4793,
      "step": 1440
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019809806295067518,
      "loss": 2.4497,
      "step": 1450
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019806987307455386,
      "loss": 2.4673,
      "step": 1460
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019804147785745365,
      "loss": 2.4531,
      "step": 1470
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019801287735882902,
      "loss": 2.4702,
      "step": 1480
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019798407163856445,
      "loss": 2.4723,
      "step": 1490
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019795506075697395,
      "loss": 2.4641,
      "step": 1500
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019792584477480122,
      "loss": 2.4383,
      "step": 1510
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019789642375321934,
      "loss": 2.4435,
      "step": 1520
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019786679775383073,
      "loss": 2.4815,
      "step": 1530
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019783696683866702,
      "loss": 2.459,
      "step": 1540
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019780693107018883,
      "loss": 2.4581,
      "step": 1550
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019777669051128583,
      "loss": 2.476,
      "step": 1560
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019774624522527634,
      "loss": 2.465,
      "step": 1570
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0001977155952759074,
      "loss": 2.4681,
      "step": 1580
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019768474072735462,
      "loss": 2.4557,
      "step": 1590
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019765368164422194,
      "loss": 2.4589,
      "step": 1600
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019762241809154164,
      "loss": 2.4721,
      "step": 1610
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.000197590950134774,
      "loss": 2.4358,
      "step": 1620
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019755927783980743,
      "loss": 2.4473,
      "step": 1630
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019752740127295808,
      "loss": 2.4597,
      "step": 1640
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019749532050096987,
      "loss": 2.4635,
      "step": 1650
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019746303559101423,
      "loss": 2.4613,
      "step": 1660
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019743054661069005,
      "loss": 2.434,
      "step": 1670
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019739785362802355,
      "loss": 2.4583,
      "step": 1680
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019736495671146805,
      "loss": 2.4445,
      "step": 1690
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0001973318559299039,
      "loss": 2.4526,
      "step": 1700
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019729855135263823,
      "loss": 2.4444,
      "step": 1710
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019726504304940496,
      "loss": 2.4348,
      "step": 1720
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019723133109036457,
      "loss": 2.4788,
      "step": 1730
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0001971974155461039,
      "loss": 2.4486,
      "step": 1740
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019716329648763616,
      "loss": 2.4431,
      "step": 1750
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019712897398640062,
      "loss": 2.4576,
      "step": 1760
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019709444811426247,
      "loss": 2.4411,
      "step": 1770
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019705971894351288,
      "loss": 2.4575,
      "step": 1780
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019702478654686848,
      "loss": 2.438,
      "step": 1790
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019698965099747165,
      "loss": 2.4322,
      "step": 1800
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019695431236888992,
      "loss": 2.4797,
      "step": 1810
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0001969187707351162,
      "loss": 2.4608,
      "step": 1820
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019688302617056836,
      "loss": 2.4586,
      "step": 1830
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0001968470787500892,
      "loss": 2.4422,
      "step": 1840
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0001968109285489463,
      "loss": 2.4493,
      "step": 1850
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019677457564283172,
      "loss": 2.4497,
      "step": 1860
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019673802010786208,
      "loss": 2.4427,
      "step": 1870
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0001967012620205782,
      "loss": 2.4482,
      "step": 1880
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019666430145794496,
      "loss": 2.4277,
      "step": 1890
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019662713849735134,
      "loss": 2.4619,
      "step": 1900
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019658977321660993,
      "loss": 2.4354,
      "step": 1910
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0001965522056939571,
      "loss": 2.458,
      "step": 1920
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0001965144360080525,
      "loss": 2.458,
      "step": 1930
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019647646423797926,
      "loss": 2.4167,
      "step": 1940
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019643829046324352,
      "loss": 2.4658,
      "step": 1950
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019639991476377443,
      "loss": 2.4474,
      "step": 1960
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019636133721992387,
      "loss": 2.4429,
      "step": 1970
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019632255791246646,
      "loss": 2.4336,
      "step": 1980
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019628357692259924,
      "loss": 2.4487,
      "step": 1990
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019624439433194144,
      "loss": 2.4497,
      "step": 2000
    },
    {
      "epoch": 0.09,
      "eval_accuracy": 0.5302039425781068,
      "eval_loss": 2.2889890670776367,
      "eval_runtime": 10.496,
      "eval_samples_per_second": 125.477,
      "eval_steps_per_second": 1.048,
      "step": 2000
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0001962050102225345,
      "loss": 2.4419,
      "step": 2010
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0001961654246768418,
      "loss": 2.4419,
      "step": 2020
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019612563777774846,
      "loss": 2.4574,
      "step": 2030
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019608564960856123,
      "loss": 2.4221,
      "step": 2040
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019604546025300828,
      "loss": 2.4448,
      "step": 2050
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019600506979523897,
      "loss": 2.425,
      "step": 2060
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0001959644783198238,
      "loss": 2.459,
      "step": 2070
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019592368591175415,
      "loss": 2.4464,
      "step": 2080
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0001958826926564421,
      "loss": 2.4366,
      "step": 2090
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0001958414986397203,
      "loss": 2.4411,
      "step": 2100
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019580010394784172,
      "loss": 2.4026,
      "step": 2110
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019575850866747953,
      "loss": 2.4511,
      "step": 2120
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019571671288572693,
      "loss": 2.4216,
      "step": 2130
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019567471669009685,
      "loss": 2.4393,
      "step": 2140
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019563252016852194,
      "loss": 2.4549,
      "step": 2150
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0001955901234093542,
      "loss": 2.4517,
      "step": 2160
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.000195547526501365,
      "loss": 2.4431,
      "step": 2170
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019550472953374474,
      "loss": 2.4349,
      "step": 2180
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0001954617325961027,
      "loss": 2.4337,
      "step": 2190
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019541853577846686,
      "loss": 2.4197,
      "step": 2200
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019537513917128367,
      "loss": 2.4508,
      "step": 2210
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019533154286541801,
      "loss": 2.4115,
      "step": 2220
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019528774695215287,
      "loss": 2.4269,
      "step": 2230
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019524375152318909,
      "loss": 2.4323,
      "step": 2240
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0001951995566706453,
      "loss": 2.4323,
      "step": 2250
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0001951551624870578,
      "loss": 2.4426,
      "step": 2260
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0001951105690653801,
      "loss": 2.4389,
      "step": 2270
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019506577649898294,
      "loss": 2.448,
      "step": 2280
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0001950207848816541,
      "loss": 2.4289,
      "step": 2290
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019497559430759796,
      "loss": 2.4392,
      "step": 2300
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019493020487143569,
      "loss": 2.437,
      "step": 2310
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0001948846166682047,
      "loss": 2.4371,
      "step": 2320
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019483882979335867,
      "loss": 2.4171,
      "step": 2330
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0001947928443427672,
      "loss": 2.4147,
      "step": 2340
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019474666041271568,
      "loss": 2.4345,
      "step": 2350
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019470027809990513,
      "loss": 2.4404,
      "step": 2360
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019465369750145187,
      "loss": 2.454,
      "step": 2370
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019460691871488742,
      "loss": 2.4344,
      "step": 2380
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0001945599418381584,
      "loss": 2.4523,
      "step": 2390
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.000194512766969626,
      "loss": 2.431,
      "step": 2400
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019446539420806604,
      "loss": 2.4223,
      "step": 2410
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019441782365266878,
      "loss": 2.4584,
      "step": 2420
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0001943700554030385,
      "loss": 2.4183,
      "step": 2430
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0001943220895591935,
      "loss": 2.4164,
      "step": 2440
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019427392622156577,
      "loss": 2.4048,
      "step": 2450
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0001942255654910008,
      "loss": 2.4366,
      "step": 2460
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019417700746875743,
      "loss": 2.4395,
      "step": 2470
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0001941282522565076,
      "loss": 2.4041,
      "step": 2480
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019407929995633608,
      "loss": 2.4511,
      "step": 2490
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019403015067074035,
      "loss": 2.3915,
      "step": 2500
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0001939808045026303,
      "loss": 2.4129,
      "step": 2510
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0001939312615553281,
      "loss": 2.4414,
      "step": 2520
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0001938815219325679,
      "loss": 2.4143,
      "step": 2530
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0001938315857384957,
      "loss": 2.4359,
      "step": 2540
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0001937814530776691,
      "loss": 2.4252,
      "step": 2550
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.000193731124055057,
      "loss": 2.4249,
      "step": 2560
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019368059877603941,
      "loss": 2.4223,
      "step": 2570
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019362987734640745,
      "loss": 2.4116,
      "step": 2580
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019357895987236274,
      "loss": 2.4108,
      "step": 2590
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019352784646051752,
      "loss": 2.4274,
      "step": 2600
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0001934765372178942,
      "loss": 2.4025,
      "step": 2610
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0001934250322519253,
      "loss": 2.4278,
      "step": 2620
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019337333167045309,
      "loss": 2.4119,
      "step": 2630
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019332143558172947,
      "loss": 2.4189,
      "step": 2640
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019326934409441565,
      "loss": 2.4447,
      "step": 2650
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019321705731758203,
      "loss": 2.4089,
      "step": 2660
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019316457536070787,
      "loss": 2.4294,
      "step": 2670
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019311189833368106,
      "loss": 2.4435,
      "step": 2680
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019305902634679804,
      "loss": 2.4375,
      "step": 2690
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019300595951076338,
      "loss": 2.4381,
      "step": 2700
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0001929526979366897,
      "loss": 2.4145,
      "step": 2710
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019289924173609727,
      "loss": 2.4238,
      "step": 2720
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0001928455910209139,
      "loss": 2.4317,
      "step": 2730
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0001927917459034748,
      "loss": 2.4323,
      "step": 2740
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019273770649652202,
      "loss": 2.4067,
      "step": 2750
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019268347291320458,
      "loss": 2.4032,
      "step": 2760
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019262904526707805,
      "loss": 2.4418,
      "step": 2770
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019257442367210425,
      "loss": 2.4239,
      "step": 2780
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019251960824265117,
      "loss": 2.4211,
      "step": 2790
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019246459909349258,
      "loss": 2.4137,
      "step": 2800
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019240939633980797,
      "loss": 2.4102,
      "step": 2810
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019235400009718213,
      "loss": 2.4359,
      "step": 2820
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.000192298410481605,
      "loss": 2.4166,
      "step": 2830
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019224262760947144,
      "loss": 2.4513,
      "step": 2840
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019218665159758086,
      "loss": 2.4078,
      "step": 2850
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0001921304825631372,
      "loss": 2.4249,
      "step": 2860
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019207412062374843,
      "loss": 2.4068,
      "step": 2870
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019201756589742652,
      "loss": 2.4054,
      "step": 2880
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019196081850258708,
      "loss": 2.4044,
      "step": 2890
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0001919038785580491,
      "loss": 2.4108,
      "step": 2900
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019184674618303476,
      "loss": 2.4189,
      "step": 2910
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019178942149716916,
      "loss": 2.4104,
      "step": 2920
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019173190462048008,
      "loss": 2.418,
      "step": 2930
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019167419567339767,
      "loss": 2.4112,
      "step": 2940
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019161629477675428,
      "loss": 2.4178,
      "step": 2950
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0001915582020517841,
      "loss": 2.4017,
      "step": 2960
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019149991762012312,
      "loss": 2.4271,
      "step": 2970
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019144144160380857,
      "loss": 2.3857,
      "step": 2980
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019138277412527889,
      "loss": 2.4088,
      "step": 2990
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019132391530737343,
      "loss": 2.386,
      "step": 3000
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019126486527333217,
      "loss": 2.4218,
      "step": 3010
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0001912056241467954,
      "loss": 2.4485,
      "step": 3020
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0001911461920518036,
      "loss": 2.4105,
      "step": 3030
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019108656911279707,
      "loss": 2.4086,
      "step": 3040
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019102675545461568,
      "loss": 2.411,
      "step": 3050
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019096675120249868,
      "loss": 2.4159,
      "step": 3060
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019090655648208436,
      "loss": 2.4106,
      "step": 3070
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0001908461714194098,
      "loss": 2.4177,
      "step": 3080
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0001907855961409107,
      "loss": 2.4199,
      "step": 3090
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019072483077342098,
      "loss": 2.4028,
      "step": 3100
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019066387544417252,
      "loss": 2.4306,
      "step": 3110
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019060273028079507,
      "loss": 2.3988,
      "step": 3120
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019054139541131571,
      "loss": 2.4424,
      "step": 3130
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0001904798709641589,
      "loss": 2.4114,
      "step": 3140
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019041815706814593,
      "loss": 2.3974,
      "step": 3150
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0001903562538524948,
      "loss": 2.3989,
      "step": 3160
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0001902941614468199,
      "loss": 2.4268,
      "step": 3170
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0001902318799811317,
      "loss": 2.4028,
      "step": 3180
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00019016940958583662,
      "loss": 2.4208,
      "step": 3190
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00019010675039173665,
      "loss": 2.3983,
      "step": 3200
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.000190043902530029,
      "loss": 2.4205,
      "step": 3210
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00018998086613230606,
      "loss": 2.4077,
      "step": 3220
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00018991764133055486,
      "loss": 2.4044,
      "step": 3230
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00018985422825715692,
      "loss": 2.3765,
      "step": 3240
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00018979062704488814,
      "loss": 2.4315,
      "step": 3250
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00018972683782691804,
      "loss": 2.4289,
      "step": 3260
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00018966286073681012,
      "loss": 2.4033,
      "step": 3270
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00018959869590852102,
      "loss": 2.3982,
      "step": 3280
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00018953434347640055,
      "loss": 2.4293,
      "step": 3290
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00018946980357519136,
      "loss": 2.4087,
      "step": 3300
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00018940507634002853,
      "loss": 2.412,
      "step": 3310
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00018934016190643954,
      "loss": 2.414,
      "step": 3320
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00018927506041034367,
      "loss": 2.4216,
      "step": 3330
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00018920977198805199,
      "loss": 2.4053,
      "step": 3340
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00018914429677626687,
      "loss": 2.4182,
      "step": 3350
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00018907863491208183,
      "loss": 2.4077,
      "step": 3360
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00018901278653298128,
      "loss": 2.4089,
      "step": 3370
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00018894675177684003,
      "loss": 2.4061,
      "step": 3380
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00018888053078192315,
      "loss": 2.4029,
      "step": 3390
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0001888141236868858,
      "loss": 2.4107,
      "step": 3400
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00018874753063077263,
      "loss": 2.4016,
      "step": 3410
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00018868075175301768,
      "loss": 2.4213,
      "step": 3420
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00018861378719344424,
      "loss": 2.401,
      "step": 3430
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00018854663709226416,
      "loss": 2.4078,
      "step": 3440
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0001884793015900779,
      "loss": 2.4128,
      "step": 3450
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00018841178082787416,
      "loss": 2.3948,
      "step": 3460
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00018834407494702942,
      "loss": 2.4056,
      "step": 3470
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00018827618408930781,
      "loss": 2.3651,
      "step": 3480
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00018820810839686086,
      "loss": 2.3947,
      "step": 3490
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.000188139848012227,
      "loss": 2.3893,
      "step": 3500
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0001880714030783314,
      "loss": 2.4021,
      "step": 3510
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00018800277373848563,
      "loss": 2.4101,
      "step": 3520
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00018793396013638743,
      "loss": 2.4058,
      "step": 3530
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00018786496241612034,
      "loss": 2.3911,
      "step": 3540
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0001877957807221533,
      "loss": 2.4102,
      "step": 3550
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00018772641519934066,
      "loss": 2.4375,
      "step": 3560
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00018765686599292152,
      "loss": 2.4198,
      "step": 3570
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00018758713324851963,
      "loss": 2.4036,
      "step": 3580
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.000187517217112143,
      "loss": 2.3893,
      "step": 3590
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00018744711773018366,
      "loss": 2.3959,
      "step": 3600
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00018737683524941735,
      "loss": 2.393,
      "step": 3610
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00018730636981700318,
      "loss": 2.3963,
      "step": 3620
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00018723572158048328,
      "loss": 2.4216,
      "step": 3630
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00018716489068778256,
      "loss": 2.4168,
      "step": 3640
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0001870938772872084,
      "loss": 2.4178,
      "step": 3650
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00018702268152745032,
      "loss": 2.4108,
      "step": 3660
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00018695130355757966,
      "loss": 2.3884,
      "step": 3670
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00018687974352704924,
      "loss": 2.3969,
      "step": 3680
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00018680800158569317,
      "loss": 2.393,
      "step": 3690
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00018673607788372633,
      "loss": 2.426,
      "step": 3700
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00018666397257174428,
      "loss": 2.4089,
      "step": 3710
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00018659168580072275,
      "loss": 2.3972,
      "step": 3720
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00018651921772201753,
      "loss": 2.3956,
      "step": 3730
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00018644656848736392,
      "loss": 2.3882,
      "step": 3740
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0001863737382488766,
      "loss": 2.3841,
      "step": 3750
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00018630072715904918,
      "loss": 2.4086,
      "step": 3760
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.000186227535370754,
      "loss": 2.4197,
      "step": 3770
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00018615416303724178,
      "loss": 2.4327,
      "step": 3780
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0001860806103121411,
      "loss": 2.408,
      "step": 3790
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00018600687734945844,
      "loss": 2.3878,
      "step": 3800
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00018593296430357762,
      "loss": 2.3852,
      "step": 3810
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0001858588713292594,
      "loss": 2.3824,
      "step": 3820
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00018578459858164146,
      "loss": 2.3784,
      "step": 3830
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00018571014621623775,
      "loss": 2.3991,
      "step": 3840
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00018563551438893837,
      "loss": 2.409,
      "step": 3850
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00018556070325600918,
      "loss": 2.3838,
      "step": 3860
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00018548571297409149,
      "loss": 2.3909,
      "step": 3870
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00018541054370020167,
      "loss": 2.3895,
      "step": 3880
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00018533519559173092,
      "loss": 2.4281,
      "step": 3890
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00018525966880644483,
      "loss": 2.4061,
      "step": 3900
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00018518396350248313,
      "loss": 2.3847,
      "step": 3910
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00018510807983835936,
      "loss": 2.4062,
      "step": 3920
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0001850320179729605,
      "loss": 2.4095,
      "step": 3930
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0001849557780655467,
      "loss": 2.3892,
      "step": 3940
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00018487936027575076,
      "loss": 2.3906,
      "step": 3950
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00018480276476357806,
      "loss": 2.3867,
      "step": 3960
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0001847259916894061,
      "loss": 2.3978,
      "step": 3970
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00018464904121398408,
      "loss": 2.412,
      "step": 3980
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0001845719134984327,
      "loss": 2.4178,
      "step": 3990
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00018449460870424377,
      "loss": 2.4021,
      "step": 4000
    },
    {
      "epoch": 0.18,
      "eval_accuracy": 0.5356110892153217,
      "eval_loss": 2.2496390342712402,
      "eval_runtime": 10.5309,
      "eval_samples_per_second": 125.061,
      "eval_steps_per_second": 1.045,
      "step": 4000
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00018441712699327989,
      "loss": 2.3893,
      "step": 4010
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00018433946852777403,
      "loss": 2.373,
      "step": 4020
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0001842616334703293,
      "loss": 2.3985,
      "step": 4030
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00018418362198391853,
      "loss": 2.3982,
      "step": 4040
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00018410543423188404,
      "loss": 2.3929,
      "step": 4050
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0001840270703779371,
      "loss": 2.3961,
      "step": 4060
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00018394853058615788,
      "loss": 2.4061,
      "step": 4070
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00018386981502099465,
      "loss": 2.4216,
      "step": 4080
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.000183790923847264,
      "loss": 2.4063,
      "step": 4090
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00018371185723015007,
      "loss": 2.3886,
      "step": 4100
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00018363261533520443,
      "loss": 2.3979,
      "step": 4110
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0001835531983283455,
      "loss": 2.3898,
      "step": 4120
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00018347360637585857,
      "loss": 2.404,
      "step": 4130
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00018339383964439504,
      "loss": 2.3835,
      "step": 4140
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0001833138983009724,
      "loss": 2.3896,
      "step": 4150
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0001832337825129736,
      "loss": 2.3826,
      "step": 4160
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00018315349244814707,
      "loss": 2.4093,
      "step": 4170
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0001830730282746059,
      "loss": 2.3962,
      "step": 4180
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00018299239016082792,
      "loss": 2.3983,
      "step": 4190
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00018291157827565505,
      "loss": 2.3884,
      "step": 4200
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0001828305927882931,
      "loss": 2.3952,
      "step": 4210
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00018274943386831135,
      "loss": 2.3872,
      "step": 4220
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00018266810168564223,
      "loss": 2.4018,
      "step": 4230
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.000182586596410581,
      "loss": 2.3948,
      "step": 4240
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00018250491821378525,
      "loss": 2.3805,
      "step": 4250
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00018242306726627474,
      "loss": 2.3928,
      "step": 4260
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00018234104373943084,
      "loss": 2.3957,
      "step": 4270
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00018225884780499632,
      "loss": 2.4042,
      "step": 4280
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00018217647963507494,
      "loss": 2.3872,
      "step": 4290
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0001820939394021311,
      "loss": 2.3861,
      "step": 4300
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00018201122727898948,
      "loss": 2.4026,
      "step": 4310
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0001819283434388346,
      "loss": 2.3895,
      "step": 4320
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00018184528805521055,
      "loss": 2.3909,
      "step": 4330
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0001817620613020207,
      "loss": 2.3991,
      "step": 4340
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0001816786633535271,
      "loss": 2.3907,
      "step": 4350
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00018159509438435035,
      "loss": 2.3951,
      "step": 4360
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00018151135456946903,
      "loss": 2.3823,
      "step": 4370
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0001814274440842196,
      "loss": 2.4021,
      "step": 4380
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00018134336310429575,
      "loss": 2.3981,
      "step": 4390
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0001812591118057482,
      "loss": 2.4157,
      "step": 4400
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0001811746903649842,
      "loss": 2.3981,
      "step": 4410
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00018109009895876744,
      "loss": 2.3924,
      "step": 4420
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00018100533776421732,
      "loss": 2.3973,
      "step": 4430
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00018092040695880877,
      "loss": 2.3862,
      "step": 4440
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0001808353067203719,
      "loss": 2.3788,
      "step": 4450
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00018075003722709156,
      "loss": 2.3876,
      "step": 4460
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00018066459865750706,
      "loss": 2.3864,
      "step": 4470
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00018057899119051162,
      "loss": 2.381,
      "step": 4480
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00018049321500535209,
      "loss": 2.3756,
      "step": 4490
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00018040727028162873,
      "loss": 2.3846,
      "step": 4500
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0001803211571992945,
      "loss": 2.3926,
      "step": 4510
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00018023487593865506,
      "loss": 2.3968,
      "step": 4520
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00018014842668036808,
      "loss": 2.3883,
      "step": 4530
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00018006180960544307,
      "loss": 2.379,
      "step": 4540
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00017997502489524081,
      "loss": 2.3949,
      "step": 4550
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00017988807273147317,
      "loss": 2.3997,
      "step": 4560
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00017980095329620262,
      "loss": 2.3659,
      "step": 4570
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00017971366677184183,
      "loss": 2.4092,
      "step": 4580
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0001796262133411534,
      "loss": 2.3843,
      "step": 4590
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0001795385931872493,
      "loss": 2.3884,
      "step": 4600
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00017945080649359069,
      "loss": 2.373,
      "step": 4610
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00017936285344398732,
      "loss": 2.3992,
      "step": 4620
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00017927473422259743,
      "loss": 2.3743,
      "step": 4630
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00017918644901392705,
      "loss": 2.389,
      "step": 4640
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00017909799800282976,
      "loss": 2.3865,
      "step": 4650
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00017900938137450638,
      "loss": 2.3862,
      "step": 4660
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00017892059931450446,
      "loss": 2.4079,
      "step": 4670
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00017883165200871797,
      "loss": 2.4191,
      "step": 4680
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0001787425396433868,
      "loss": 2.3877,
      "step": 4690
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0001786532624050965,
      "loss": 2.4038,
      "step": 4700
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00017856382048077782,
      "loss": 2.4007,
      "step": 4710
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0001784742140577064,
      "loss": 2.3739,
      "step": 4720
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00017838444332350214,
      "loss": 2.3765,
      "step": 4730
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0001782945084661292,
      "loss": 2.3829,
      "step": 4740
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0001782044096738952,
      "loss": 2.3895,
      "step": 4750
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00017811414713545114,
      "loss": 2.3742,
      "step": 4760
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00017802372103979078,
      "loss": 2.3856,
      "step": 4770
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00017793313157625038,
      "loss": 2.3929,
      "step": 4780
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00017784237893450828,
      "loss": 2.3837,
      "step": 4790
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00017775146330458448,
      "loss": 2.3861,
      "step": 4800
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0001776603848768402,
      "loss": 2.387,
      "step": 4810
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00017756914384197762,
      "loss": 2.3694,
      "step": 4820
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00017747774039103927,
      "loss": 2.3548,
      "step": 4830
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00017738617471540788,
      "loss": 2.366,
      "step": 4840
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00017729444700680577,
      "loss": 2.4099,
      "step": 4850
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00017720255745729457,
      "loss": 2.3567,
      "step": 4860
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0001771105062592747,
      "loss": 2.4122,
      "step": 4870
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00017701829360548517,
      "loss": 2.3988,
      "step": 4880
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00017692591968900288,
      "loss": 2.3775,
      "step": 4890
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0001768333847032426,
      "loss": 2.3915,
      "step": 4900
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00017674068884195612,
      "loss": 2.4085,
      "step": 4910
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00017664783229923226,
      "loss": 2.3651,
      "step": 4920
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00017655481526949616,
      "loss": 2.3697,
      "step": 4930
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0001764616379475091,
      "loss": 2.3695,
      "step": 4940
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0001763683005283679,
      "loss": 2.3927,
      "step": 4950
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00017627480320750455,
      "loss": 2.4027,
      "step": 4960
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00017618114618068596,
      "loss": 2.3786,
      "step": 4970
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00017608732964401343,
      "loss": 2.3923,
      "step": 4980
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00017599335379392219,
      "loss": 2.3938,
      "step": 4990
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00017589921882718102,
      "loss": 2.3768,
      "step": 5000
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00017580492494089192,
      "loss": 2.3899,
      "step": 5010
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00017571047233248962,
      "loss": 2.4152,
      "step": 5020
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00017561586119974116,
      "loss": 2.4003,
      "step": 5030
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00017552109174074555,
      "loss": 2.3835,
      "step": 5040
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00017542616415393332,
      "loss": 2.3748,
      "step": 5050
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.000175331078638066,
      "loss": 2.402,
      "step": 5060
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00017523583539223587,
      "loss": 2.3572,
      "step": 5070
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0001751404346158655,
      "loss": 2.3714,
      "step": 5080
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0001750448765087072,
      "loss": 2.3893,
      "step": 5090
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00017494916127084274,
      "loss": 2.3862,
      "step": 5100
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00017485328910268292,
      "loss": 2.3715,
      "step": 5110
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00017475726020496719,
      "loss": 2.4044,
      "step": 5120
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00017466107477876296,
      "loss": 2.3896,
      "step": 5130
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00017456473302546562,
      "loss": 2.3748,
      "step": 5140
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0001744682351467977,
      "loss": 2.3865,
      "step": 5150
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00017437158134480876,
      "loss": 2.3701,
      "step": 5160
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00017427477182187473,
      "loss": 2.365,
      "step": 5170
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00017417780678069764,
      "loss": 2.3659,
      "step": 5180
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00017408068642430516,
      "loss": 2.3822,
      "step": 5190
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00017398341095605017,
      "loss": 2.3841,
      "step": 5200
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00017388598057961025,
      "loss": 2.3834,
      "step": 5210
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00017378839549898745,
      "loss": 2.4141,
      "step": 5220
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00017369065591850763,
      "loss": 2.3674,
      "step": 5230
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00017359276204282023,
      "loss": 2.3714,
      "step": 5240
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0001734947140768977,
      "loss": 2.379,
      "step": 5250
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0001733965122260351,
      "loss": 2.3895,
      "step": 5260
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0001732981566958499,
      "loss": 2.373,
      "step": 5270
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00017319964769228104,
      "loss": 2.372,
      "step": 5280
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00017310098542158903,
      "loss": 2.3674,
      "step": 5290
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0001730021700903552,
      "loss": 2.3651,
      "step": 5300
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0001729032019054814,
      "loss": 2.3757,
      "step": 5310
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0001728040810741895,
      "loss": 2.379,
      "step": 5320
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00017270480780402103,
      "loss": 2.3587,
      "step": 5330
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00017260538230283668,
      "loss": 2.3783,
      "step": 5340
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0001725058047788158,
      "loss": 2.3608,
      "step": 5350
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00017240607544045618,
      "loss": 2.3824,
      "step": 5360
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00017230619449657341,
      "loss": 2.4014,
      "step": 5370
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00017220616215630056,
      "loss": 2.3696,
      "step": 5380
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00017210597862908763,
      "loss": 2.3645,
      "step": 5390
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00017200564412470123,
      "loss": 2.3684,
      "step": 5400
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0001719051588532241,
      "loss": 2.3615,
      "step": 5410
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00017180452302505462,
      "loss": 2.3866,
      "step": 5420
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00017170373685090637,
      "loss": 2.3883,
      "step": 5430
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0001716028005418079,
      "loss": 2.3795,
      "step": 5440
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00017150171430910187,
      "loss": 2.3643,
      "step": 5450
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00017140047836444508,
      "loss": 2.377,
      "step": 5460
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00017129909291980767,
      "loss": 2.3633,
      "step": 5470
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00017119755818747283,
      "loss": 2.3656,
      "step": 5480
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00017109587438003638,
      "loss": 2.4058,
      "step": 5490
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00017099404171040618,
      "loss": 2.3744,
      "step": 5500
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00017089206039180187,
      "loss": 2.3962,
      "step": 5510
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00017078993063775433,
      "loss": 2.3937,
      "step": 5520
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00017068765266210515,
      "loss": 2.3879,
      "step": 5530
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00017058522667900638,
      "loss": 2.3537,
      "step": 5540
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00017048265290291992,
      "loss": 2.3575,
      "step": 5550
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00017037993154861712,
      "loss": 2.3659,
      "step": 5560
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0001702770628311783,
      "loss": 2.3788,
      "step": 5570
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00017017404696599244,
      "loss": 2.3753,
      "step": 5580
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00017007088416875652,
      "loss": 2.3699,
      "step": 5590
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00016996757465547518,
      "loss": 2.3619,
      "step": 5600
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0001698641186424603,
      "loss": 2.3943,
      "step": 5610
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0001697605163463305,
      "loss": 2.3801,
      "step": 5620
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00016965676798401066,
      "loss": 2.3857,
      "step": 5630
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0001695528737727315,
      "loss": 2.3686,
      "step": 5640
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00016944883393002914,
      "loss": 2.3527,
      "step": 5650
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0001693446486737446,
      "loss": 2.3712,
      "step": 5660
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00016924031822202347,
      "loss": 2.3771,
      "step": 5670
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00016913584279331517,
      "loss": 2.3701,
      "step": 5680
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00016903122260637288,
      "loss": 2.4011,
      "step": 5690
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00016892645788025266,
      "loss": 2.3826,
      "step": 5700
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00016882154883431346,
      "loss": 2.3886,
      "step": 5710
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00016871649568821616,
      "loss": 2.3818,
      "step": 5720
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00016861129866192355,
      "loss": 2.3776,
      "step": 5730
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00016850595797569958,
      "loss": 2.3581,
      "step": 5740
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00016840047385010905,
      "loss": 2.3507,
      "step": 5750
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00016829484650601707,
      "loss": 2.3819,
      "step": 5760
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0001681890761645886,
      "loss": 2.383,
      "step": 5770
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00016808316304728804,
      "loss": 2.3882,
      "step": 5780
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00016797710737587879,
      "loss": 2.3791,
      "step": 5790
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00016787090937242258,
      "loss": 2.3823,
      "step": 5800
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00016776456925927935,
      "loss": 2.3603,
      "step": 5810
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00016765808725910646,
      "loss": 2.3664,
      "step": 5820
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00016755146359485838,
      "loss": 2.3645,
      "step": 5830
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0001674446984897863,
      "loss": 2.3769,
      "step": 5840
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00016733779216743737,
      "loss": 2.3579,
      "step": 5850
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00016723074485165457,
      "loss": 2.3575,
      "step": 5860
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0001671235567665761,
      "loss": 2.3735,
      "step": 5870
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00016701622813663477,
      "loss": 2.3754,
      "step": 5880
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00016690875918655787,
      "loss": 2.3559,
      "step": 5890
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0001668011501413663,
      "loss": 2.3643,
      "step": 5900
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00016669340122637443,
      "loss": 2.3802,
      "step": 5910
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0001665855126671894,
      "loss": 2.3672,
      "step": 5920
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0001664774846897108,
      "loss": 2.3804,
      "step": 5930
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00016636931752013018,
      "loss": 2.3821,
      "step": 5940
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00016626101138493036,
      "loss": 2.3484,
      "step": 5950
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0001661525665108853,
      "loss": 2.395,
      "step": 5960
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0001660439831250594,
      "loss": 2.3502,
      "step": 5970
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00016593526145480708,
      "loss": 2.3561,
      "step": 5980
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00016582640172777233,
      "loss": 2.3821,
      "step": 5990
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0001657174041718881,
      "loss": 2.3883,
      "step": 6000
    },
    {
      "epoch": 0.28,
      "eval_accuracy": 0.5390075343782449,
      "eval_loss": 2.2250747680664062,
      "eval_runtime": 10.5596,
      "eval_samples_per_second": 124.721,
      "eval_steps_per_second": 1.042,
      "step": 6000
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00016560826901537606,
      "loss": 2.3956,
      "step": 6010
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00016549899648674588,
      "loss": 2.3826,
      "step": 6020
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.000165389586814795,
      "loss": 2.3777,
      "step": 6030
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00016528004022860787,
      "loss": 2.3843,
      "step": 6040
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00016517035695755568,
      "loss": 2.3767,
      "step": 6050
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00016506053723129588,
      "loss": 2.355,
      "step": 6060
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0001649505812797715,
      "loss": 2.3792,
      "step": 6070
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00016484048933321086,
      "loss": 2.3639,
      "step": 6080
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00016473026162212707,
      "loss": 2.3617,
      "step": 6090
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00016461989837731746,
      "loss": 2.3728,
      "step": 6100
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00016450939982986314,
      "loss": 2.3831,
      "step": 6110
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0001643987662111286,
      "loss": 2.3744,
      "step": 6120
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00016428799775276097,
      "loss": 2.3583,
      "step": 6130
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00016417709468668994,
      "loss": 2.3703,
      "step": 6140
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00016406605724512684,
      "loss": 2.3688,
      "step": 6150
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00016395488566056448,
      "loss": 2.3695,
      "step": 6160
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0001638435801657765,
      "loss": 2.381,
      "step": 6170
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00016373214099381695,
      "loss": 2.3648,
      "step": 6180
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00016362056837801973,
      "loss": 2.3556,
      "step": 6190
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00016350886255199823,
      "loss": 2.3627,
      "step": 6200
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0001633970237496446,
      "loss": 2.3685,
      "step": 6210
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00016328505220512964,
      "loss": 2.3835,
      "step": 6220
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0001631729481529019,
      "loss": 2.3625,
      "step": 6230
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00016306071182768745,
      "loss": 2.3655,
      "step": 6240
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00016294834346448935,
      "loss": 2.3675,
      "step": 6250
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00016283584329858708,
      "loss": 2.3776,
      "step": 6260
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00016272321156553604,
      "loss": 2.3632,
      "step": 6270
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0001626104485011673,
      "loss": 2.3765,
      "step": 6280
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00016249755434158663,
      "loss": 2.3878,
      "step": 6290
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00016238452932317458,
      "loss": 2.3661,
      "step": 6300
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00016227137368258546,
      "loss": 2.3566,
      "step": 6310
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00016215808765674726,
      "loss": 2.3554,
      "step": 6320
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00016204467148286082,
      "loss": 2.3598,
      "step": 6330
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00016193112539839962,
      "loss": 2.3763,
      "step": 6340
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00016181744964110904,
      "loss": 2.3623,
      "step": 6350
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0001617036444490061,
      "loss": 2.3628,
      "step": 6360
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00016158971006037865,
      "loss": 2.3678,
      "step": 6370
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0001614756467137852,
      "loss": 2.3427,
      "step": 6380
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0001613614546480543,
      "loss": 2.3687,
      "step": 6390
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00016124713410228388,
      "loss": 2.3534,
      "step": 6400
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00016113268531584094,
      "loss": 2.3498,
      "step": 6410
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00016101810852836104,
      "loss": 2.3594,
      "step": 6420
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00016090340397974777,
      "loss": 2.3492,
      "step": 6430
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00016078857191017209,
      "loss": 2.3659,
      "step": 6440
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00016067361256007212,
      "loss": 2.3693,
      "step": 6450
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0001605585261701524,
      "loss": 2.3825,
      "step": 6460
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00016044331298138345,
      "loss": 2.3372,
      "step": 6470
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00016032797323500136,
      "loss": 2.3665,
      "step": 6480
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0001602125071725072,
      "loss": 2.3535,
      "step": 6490
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0001600969150356664,
      "loss": 2.3783,
      "step": 6500
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00015998119706650855,
      "loss": 2.3785,
      "step": 6510
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0001598653535073266,
      "loss": 2.3739,
      "step": 6520
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00015974938460067647,
      "loss": 2.3809,
      "step": 6530
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00015963329058937657,
      "loss": 2.3572,
      "step": 6540
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00015951707171650721,
      "loss": 2.3462,
      "step": 6550
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00015940072822541023,
      "loss": 2.3719,
      "step": 6560
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00015928426035968825,
      "loss": 2.3473,
      "step": 6570
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00015916766836320445,
      "loss": 2.4003,
      "step": 6580
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00015905095248008183,
      "loss": 2.367,
      "step": 6590
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00015893411295470284,
      "loss": 2.373,
      "step": 6600
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00015881715003170877,
      "loss": 2.3572,
      "step": 6610
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00015870006395599933,
      "loss": 2.3404,
      "step": 6620
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00015858285497273206,
      "loss": 2.3701,
      "step": 6630
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00015846552332732187,
      "loss": 2.3648,
      "step": 6640
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00015834806926544044,
      "loss": 2.3631,
      "step": 6650
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00015823049303301584,
      "loss": 2.3427,
      "step": 6660
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00015811279487623194,
      "loss": 2.3632,
      "step": 6670
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00015799497504152786,
      "loss": 2.3597,
      "step": 6680
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00015787703377559745,
      "loss": 2.3485,
      "step": 6690
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00015775897132538894,
      "loss": 2.3606,
      "step": 6700
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00015764078793810424,
      "loss": 2.3752,
      "step": 6710
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00015752248386119842,
      "loss": 2.3457,
      "step": 6720
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00015740405934237933,
      "loss": 2.3712,
      "step": 6730
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.000157285514629607,
      "loss": 2.3453,
      "step": 6740
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00015716684997109308,
      "loss": 2.3674,
      "step": 6750
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00015704806561530038,
      "loss": 2.3543,
      "step": 6760
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00015692916181094237,
      "loss": 2.3909,
      "step": 6770
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00015681013880698258,
      "loss": 2.3788,
      "step": 6780
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00015669099685263415,
      "loss": 2.3528,
      "step": 6790
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00015657173619735928,
      "loss": 2.3626,
      "step": 6800
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0001564523570908687,
      "loss": 2.3552,
      "step": 6810
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0001563328597831211,
      "loss": 2.3638,
      "step": 6820
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00015621324452432282,
      "loss": 2.3565,
      "step": 6830
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00015609351156492697,
      "loss": 2.3564,
      "step": 6840
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0001559736611556333,
      "loss": 2.347,
      "step": 6850
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0001558536935473873,
      "loss": 2.3698,
      "step": 6860
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00015573360899138,
      "loss": 2.3543,
      "step": 6870
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00015561340773904715,
      "loss": 2.356,
      "step": 6880
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00015549309004206897,
      "loss": 2.366,
      "step": 6890
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0001553726561523695,
      "loss": 2.3477,
      "step": 6900
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00015525210632211594,
      "loss": 2.3915,
      "step": 6910
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00015513144080371834,
      "loss": 2.3613,
      "step": 6920
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00015501065984982901,
      "loss": 2.3708,
      "step": 6930
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0001548897637133419,
      "loss": 2.3831,
      "step": 6940
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00015476875264739212,
      "loss": 2.3723,
      "step": 6950
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00015464762690535548,
      "loss": 2.374,
      "step": 6960
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00015452638674084786,
      "loss": 2.3572,
      "step": 6970
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00015440503240772473,
      "loss": 2.3559,
      "step": 6980
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00015428356416008062,
      "loss": 2.3551,
      "step": 6990
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00015416198225224855,
      "loss": 2.3836,
      "step": 7000
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0001540402869387996,
      "loss": 2.3528,
      "step": 7010
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00015391847847454223,
      "loss": 2.3661,
      "step": 7020
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00015379655711452177,
      "loss": 2.3499,
      "step": 7030
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00015367452311402003,
      "loss": 2.3635,
      "step": 7040
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00015355237672855468,
      "loss": 2.3635,
      "step": 7050
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0001534301182138786,
      "loss": 2.3526,
      "step": 7060
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00015330774782597954,
      "loss": 2.3353,
      "step": 7070
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00015318526582107944,
      "loss": 2.336,
      "step": 7080
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00015306267245563396,
      "loss": 2.3942,
      "step": 7090
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.000152939967986332,
      "loss": 2.3624,
      "step": 7100
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.000152817152670095,
      "loss": 2.3426,
      "step": 7110
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00015269422676407647,
      "loss": 2.3585,
      "step": 7120
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00015257119052566157,
      "loss": 2.346,
      "step": 7130
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00015244804421246647,
      "loss": 2.376,
      "step": 7140
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0001523247880823377,
      "loss": 2.3538,
      "step": 7150
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00015220142239335188,
      "loss": 2.3514,
      "step": 7160
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0001520779474038149,
      "loss": 2.3742,
      "step": 7170
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00015195436337226163,
      "loss": 2.3508,
      "step": 7180
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00015183067055745512,
      "loss": 2.3486,
      "step": 7190
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00015170686921838633,
      "loss": 2.3559,
      "step": 7200
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0001515829596142733,
      "loss": 2.3388,
      "step": 7210
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0001514589420045609,
      "loss": 2.3687,
      "step": 7220
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00015133481664892005,
      "loss": 2.3462,
      "step": 7230
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0001512105838072473,
      "loss": 2.3717,
      "step": 7240
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00015108624373966429,
      "loss": 2.3588,
      "step": 7250
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00015096179670651713,
      "loss": 2.3789,
      "step": 7260
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00015083724296837588,
      "loss": 2.3564,
      "step": 7270
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00015071258278603412,
      "loss": 2.3616,
      "step": 7280
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00015058781642050817,
      "loss": 2.3401,
      "step": 7290
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00015046294413303678,
      "loss": 2.3366,
      "step": 7300
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00015033796618508044,
      "loss": 2.378,
      "step": 7310
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00015021288283832083,
      "loss": 2.3506,
      "step": 7320
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00015008769435466047,
      "loss": 2.376,
      "step": 7330
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0001499624009962218,
      "loss": 2.3973,
      "step": 7340
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00014983700302534709,
      "loss": 2.3523,
      "step": 7350
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00014971150070459743,
      "loss": 2.3608,
      "step": 7360
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00014958589429675255,
      "loss": 2.3578,
      "step": 7370
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00014946018406481002,
      "loss": 2.3421,
      "step": 7380
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00014933437027198486,
      "loss": 2.3525,
      "step": 7390
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00014920845318170894,
      "loss": 2.354,
      "step": 7400
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00014908243305763034,
      "loss": 2.3506,
      "step": 7410
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00014895631016361298,
      "loss": 2.3437,
      "step": 7420
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00014883008476373585,
      "loss": 2.3487,
      "step": 7430
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0001487037571222927,
      "loss": 2.3503,
      "step": 7440
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0001485773275037912,
      "loss": 2.337,
      "step": 7450
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00014845079617295268,
      "loss": 2.3461,
      "step": 7460
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00014832416339471133,
      "loss": 2.3407,
      "step": 7470
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00014819742943421383,
      "loss": 2.3621,
      "step": 7480
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00014807059455681872,
      "loss": 2.3459,
      "step": 7490
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00014794365902809578,
      "loss": 2.3394,
      "step": 7500
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00014781662311382556,
      "loss": 2.3649,
      "step": 7510
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0001476894870799988,
      "loss": 2.37,
      "step": 7520
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00014756225119281596,
      "loss": 2.3516,
      "step": 7530
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00014743491571868642,
      "loss": 2.3615,
      "step": 7540
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00014730748092422818,
      "loss": 2.3513,
      "step": 7550
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00014717994707626718,
      "loss": 2.3582,
      "step": 7560
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00014705231444183672,
      "loss": 2.3274,
      "step": 7570
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.000146924583288177,
      "loss": 2.3512,
      "step": 7580
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0001467967538827345,
      "loss": 2.341,
      "step": 7590
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00014666882649316137,
      "loss": 2.3581,
      "step": 7600
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00014654080138731495,
      "loss": 2.3414,
      "step": 7610
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0001464126788332572,
      "loss": 2.3381,
      "step": 7620
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00014628445909925408,
      "loss": 2.3557,
      "step": 7630
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00014615614245377508,
      "loss": 2.3574,
      "step": 7640
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00014602772916549254,
      "loss": 2.3636,
      "step": 7650
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0001458992195032812,
      "loss": 2.3606,
      "step": 7660
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00014577061373621756,
      "loss": 2.3495,
      "step": 7670
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00014564191213357937,
      "loss": 2.3618,
      "step": 7680
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00014551311496484507,
      "loss": 2.3608,
      "step": 7690
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00014538422249969318,
      "loss": 2.3514,
      "step": 7700
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00014525523500800164,
      "loss": 2.3364,
      "step": 7710
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00014512615275984753,
      "loss": 2.3579,
      "step": 7720
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00014499697602550623,
      "loss": 2.3583,
      "step": 7730
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00014486770507545103,
      "loss": 2.3518,
      "step": 7740
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0001447383401803524,
      "loss": 2.3524,
      "step": 7750
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0001446088816110776,
      "loss": 2.3529,
      "step": 7760
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00014447932963868997,
      "loss": 2.342,
      "step": 7770
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00014434968453444846,
      "loss": 2.347,
      "step": 7780
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00014421994656980698,
      "loss": 2.3726,
      "step": 7790
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00014409011601641388,
      "loss": 2.3629,
      "step": 7800
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00014396019314611143,
      "loss": 2.3355,
      "step": 7810
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00014383017823093515,
      "loss": 2.344,
      "step": 7820
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00014370007154311325,
      "loss": 2.3544,
      "step": 7830
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00014356987335506618,
      "loss": 2.3461,
      "step": 7840
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0001434395839394059,
      "loss": 2.3396,
      "step": 7850
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00014330920356893544,
      "loss": 2.3526,
      "step": 7860
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00014317873251664827,
      "loss": 2.3637,
      "step": 7870
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00014304817105572766,
      "loss": 2.3332,
      "step": 7880
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00014291751945954623,
      "loss": 2.3436,
      "step": 7890
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00014278677800166537,
      "loss": 2.35,
      "step": 7900
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00014265594695583454,
      "loss": 2.3459,
      "step": 7910
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00014252502659599085,
      "loss": 2.3674,
      "step": 7920
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00014239401719625837,
      "loss": 2.3521,
      "step": 7930
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00014226291903094769,
      "loss": 2.369,
      "step": 7940
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0001421317323745551,
      "loss": 2.331,
      "step": 7950
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00014200045750176234,
      "loss": 2.3348,
      "step": 7960
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00014186909468743574,
      "loss": 2.332,
      "step": 7970
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00014173764420662586,
      "loss": 2.3425,
      "step": 7980
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00014160610633456673,
      "loss": 2.3307,
      "step": 7990
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00014147448134667543,
      "loss": 2.3684,
      "step": 8000
    },
    {
      "epoch": 0.37,
      "eval_accuracy": 0.5416320601859583,
      "eval_loss": 2.2055718898773193,
      "eval_runtime": 10.6267,
      "eval_samples_per_second": 123.933,
      "eval_steps_per_second": 1.035,
      "step": 8000
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00014134276951855142,
      "loss": 2.3557,
      "step": 8010
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00014121097112597598,
      "loss": 2.3543,
      "step": 8020
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00014107908644491162,
      "loss": 2.3706,
      "step": 8030
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00014094711575150158,
      "loss": 2.3548,
      "step": 8040
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00014081505932206915,
      "loss": 2.3409,
      "step": 8050
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00014068291743311718,
      "loss": 2.3452,
      "step": 8060
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00014055069036132735,
      "loss": 2.3435,
      "step": 8070
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00014041837838355987,
      "loss": 2.3603,
      "step": 8080
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00014028598177685258,
      "loss": 2.3679,
      "step": 8090
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00014015350081842057,
      "loss": 2.3396,
      "step": 8100
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0001400209357856556,
      "loss": 2.3571,
      "step": 8110
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00013988828695612538,
      "loss": 2.3511,
      "step": 8120
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00013975555460757306,
      "loss": 2.3306,
      "step": 8130
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0001396227390179168,
      "loss": 2.3444,
      "step": 8140
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0001394898404652489,
      "loss": 2.3473,
      "step": 8150
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0001393568592278355,
      "loss": 2.3557,
      "step": 8160
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0001392237955841158,
      "loss": 2.3716,
      "step": 8170
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00013909064981270148,
      "loss": 2.3547,
      "step": 8180
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00013895742219237632,
      "loss": 2.3568,
      "step": 8190
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00013882411300209535,
      "loss": 2.339,
      "step": 8200
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0001386907225209845,
      "loss": 2.3426,
      "step": 8210
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00013855725102833983,
      "loss": 2.3802,
      "step": 8220
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00013842369880362706,
      "loss": 2.3491,
      "step": 8230
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00013829006612648098,
      "loss": 2.3382,
      "step": 8240
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00013815635327670472,
      "loss": 2.358,
      "step": 8250
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0001380225605342694,
      "loss": 2.3452,
      "step": 8260
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00013788868817931339,
      "loss": 2.3629,
      "step": 8270
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0001377547364921417,
      "loss": 2.3447,
      "step": 8280
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00013762070575322547,
      "loss": 2.341,
      "step": 8290
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0001374865962432014,
      "loss": 2.3651,
      "step": 8300
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00013735240824287113,
      "loss": 2.3603,
      "step": 8310
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00013721814203320054,
      "loss": 2.3545,
      "step": 8320
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0001370837978953194,
      "loss": 2.3543,
      "step": 8330
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00013694937611052056,
      "loss": 2.3269,
      "step": 8340
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00013681487696025953,
      "loss": 2.3518,
      "step": 8350
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00013668030072615367,
      "loss": 2.3467,
      "step": 8360
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00013654564768998188,
      "loss": 2.3542,
      "step": 8370
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00013641091813368378,
      "loss": 2.3209,
      "step": 8380
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00013627611233935929,
      "loss": 2.3351,
      "step": 8390
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0001361412305892679,
      "loss": 2.3451,
      "step": 8400
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0001360062731658281,
      "loss": 2.3416,
      "step": 8410
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00013587124035161695,
      "loss": 2.3524,
      "step": 8420
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0001357361324293693,
      "loss": 2.3534,
      "step": 8430
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0001356009496819772,
      "loss": 2.3522,
      "step": 8440
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00013546569239248946,
      "loss": 2.3458,
      "step": 8450
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00013533036084411093,
      "loss": 2.341,
      "step": 8460
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00013519495532020193,
      "loss": 2.3546,
      "step": 8470
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00013505947610427773,
      "loss": 2.3311,
      "step": 8480
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00013492392348000783,
      "loss": 2.3602,
      "step": 8490
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00013478829773121547,
      "loss": 2.3627,
      "step": 8500
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00013465259914187698,
      "loss": 2.3627,
      "step": 8510
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0001345168279961213,
      "loss": 2.3453,
      "step": 8520
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00013438098457822908,
      "loss": 2.3476,
      "step": 8530
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00013424506917263253,
      "loss": 2.342,
      "step": 8540
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00013410908206391443,
      "loss": 2.3435,
      "step": 8550
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0001339730235368078,
      "loss": 2.341,
      "step": 8560
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00013383689387619507,
      "loss": 2.3205,
      "step": 8570
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00013370069336710773,
      "loss": 2.3599,
      "step": 8580
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0001335644222947256,
      "loss": 2.3414,
      "step": 8590
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00013342808094437614,
      "loss": 2.3485,
      "step": 8600
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00013329166960153414,
      "loss": 2.3385,
      "step": 8610
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00013315518855182073,
      "loss": 2.3394,
      "step": 8620
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0001330186380810032,
      "loss": 2.3398,
      "step": 8630
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00013288201847499402,
      "loss": 2.3477,
      "step": 8640
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00013274533001985054,
      "loss": 2.3483,
      "step": 8650
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00013260857300177422,
      "loss": 2.3414,
      "step": 8660
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00013247174770711007,
      "loss": 2.3218,
      "step": 8670
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00013233485442234606,
      "loss": 2.3611,
      "step": 8680
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00013219789343411258,
      "loss": 2.3431,
      "step": 8690
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00013206086502918167,
      "loss": 2.3499,
      "step": 8700
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00013192376949446662,
      "loss": 2.3229,
      "step": 8710
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0001317866071170212,
      "loss": 2.3556,
      "step": 8720
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00013164937818403927,
      "loss": 2.3446,
      "step": 8730
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0001315120829828539,
      "loss": 2.3142,
      "step": 8740
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.000131374721800937,
      "loss": 2.3365,
      "step": 8750
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00013123729492589857,
      "loss": 2.3354,
      "step": 8760
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0001310998026454862,
      "loss": 2.3274,
      "step": 8770
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00013096224524758446,
      "loss": 2.3321,
      "step": 8780
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00013082462302021417,
      "loss": 2.3413,
      "step": 8790
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.000130686936251532,
      "loss": 2.349,
      "step": 8800
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0001305491852298297,
      "loss": 2.3373,
      "step": 8810
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00013041137024353349,
      "loss": 2.3317,
      "step": 8820
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00013027349158120367,
      "loss": 2.3229,
      "step": 8830
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00013013554953153378,
      "loss": 2.3344,
      "step": 8840
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00012999754438335005,
      "loss": 2.3531,
      "step": 8850
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0001298594764256109,
      "loss": 2.3525,
      "step": 8860
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00012972134594740618,
      "loss": 2.3227,
      "step": 8870
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00012958315323795673,
      "loss": 2.3282,
      "step": 8880
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0001294448985866137,
      "loss": 2.3536,
      "step": 8890
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0001293065822828578,
      "loss": 2.3556,
      "step": 8900
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00012916820461629896,
      "loss": 2.3073,
      "step": 8910
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00012902976587667557,
      "loss": 2.3425,
      "step": 8920
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00012889126635385389,
      "loss": 2.3223,
      "step": 8930
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00012875270633782738,
      "loss": 2.3507,
      "step": 8940
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00012861408611871628,
      "loss": 2.3441,
      "step": 8950
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0001284754059867668,
      "loss": 2.3572,
      "step": 8960
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00012833666623235065,
      "loss": 2.3241,
      "step": 8970
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00012819786714596428,
      "loss": 2.3404,
      "step": 8980
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00012805900901822852,
      "loss": 2.3476,
      "step": 8990
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00012792009213988771,
      "loss": 2.332,
      "step": 9000
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00012778111680180923,
      "loss": 2.3302,
      "step": 9010
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0001276420832949829,
      "loss": 2.3379,
      "step": 9020
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0001275029919105203,
      "loss": 2.3382,
      "step": 9030
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00012736384293965416,
      "loss": 2.34,
      "step": 9040
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00012722463667373787,
      "loss": 2.3448,
      "step": 9050
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0001270853734042448,
      "loss": 2.336,
      "step": 9060
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0001269460534227675,
      "loss": 2.353,
      "step": 9070
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00012680667702101743,
      "loss": 2.3403,
      "step": 9080
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00012666724449082416,
      "loss": 2.3312,
      "step": 9090
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00012652775612413477,
      "loss": 2.3148,
      "step": 9100
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0001263882122130132,
      "loss": 2.3232,
      "step": 9110
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00012624861304963978,
      "loss": 2.3316,
      "step": 9120
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00012610895892631042,
      "loss": 2.3495,
      "step": 9130
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00012596925013543623,
      "loss": 2.347,
      "step": 9140
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0001258294869695427,
      "loss": 2.3412,
      "step": 9150
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0001256896697212692,
      "loss": 2.3574,
      "step": 9160
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0001255497986833683,
      "loss": 2.3262,
      "step": 9170
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0001254098741487053,
      "loss": 2.3389,
      "step": 9180
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00012526989641025737,
      "loss": 2.3373,
      "step": 9190
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00012512986576111316,
      "loss": 2.3419,
      "step": 9200
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00012498978249447215,
      "loss": 2.3333,
      "step": 9210
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0001248496469036439,
      "loss": 2.3359,
      "step": 9220
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00012470945928204756,
      "loss": 2.3154,
      "step": 9230
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00012456921992321123,
      "loss": 2.3446,
      "step": 9240
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00012442892912077136,
      "loss": 2.3326,
      "step": 9250
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00012428858716847211,
      "loss": 2.3385,
      "step": 9260
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00012414819436016466,
      "loss": 2.3426,
      "step": 9270
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00012400775098980678,
      "loss": 2.3453,
      "step": 9280
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00012386725735146202,
      "loss": 2.3093,
      "step": 9290
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00012372671373929927,
      "loss": 2.3348,
      "step": 9300
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.000123586120447592,
      "loss": 2.3323,
      "step": 9310
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00012344547777071768,
      "loss": 2.3327,
      "step": 9320
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00012330478600315725,
      "loss": 2.3569,
      "step": 9330
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0001231640454394944,
      "loss": 2.3349,
      "step": 9340
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.000123023256374415,
      "loss": 2.3302,
      "step": 9350
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0001228824191027064,
      "loss": 2.3398,
      "step": 9360
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00012274153391925703,
      "loss": 2.3458,
      "step": 9370
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00012260060111905552,
      "loss": 2.3434,
      "step": 9380
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00012245962099719025,
      "loss": 2.3486,
      "step": 9390
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00012231859384884865,
      "loss": 2.3346,
      "step": 9400
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00012217751996931671,
      "loss": 2.3237,
      "step": 9410
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00012203639965397815,
      "loss": 2.3337,
      "step": 9420
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00012189523319831396,
      "loss": 2.3282,
      "step": 9430
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00012175402089790175,
      "loss": 2.3311,
      "step": 9440
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00012161276304841512,
      "loss": 2.3573,
      "step": 9450
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00012147145994562305,
      "loss": 2.3221,
      "step": 9460
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00012133011188538925,
      "loss": 2.3431,
      "step": 9470
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00012118871916367153,
      "loss": 2.3519,
      "step": 9480
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00012104728207652134,
      "loss": 2.3331,
      "step": 9490
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0001209058009200829,
      "loss": 2.3214,
      "step": 9500
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0001207642759905927,
      "loss": 2.3392,
      "step": 9510
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00012062270758437899,
      "loss": 2.339,
      "step": 9520
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00012048109599786095,
      "loss": 2.3393,
      "step": 9530
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00012033944152754823,
      "loss": 2.3193,
      "step": 9540
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00012019774447004024,
      "loss": 2.3344,
      "step": 9550
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00012005600512202557,
      "loss": 2.3298,
      "step": 9560
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00011991422378028136,
      "loss": 2.3538,
      "step": 9570
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0001197724007416727,
      "loss": 2.3357,
      "step": 9580
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00011963053630315196,
      "loss": 2.3265,
      "step": 9590
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00011948863076175816,
      "loss": 2.3273,
      "step": 9600
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00011934668441461644,
      "loss": 2.3729,
      "step": 9610
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00011920469755893738,
      "loss": 2.3381,
      "step": 9620
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00011906267049201631,
      "loss": 2.3267,
      "step": 9630
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00011892060351123287,
      "loss": 2.3404,
      "step": 9640
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00011877849691405016,
      "loss": 2.3367,
      "step": 9650
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00011863635099801431,
      "loss": 2.3371,
      "step": 9660
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0001184941660607537,
      "loss": 2.334,
      "step": 9670
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00011835194239997847,
      "loss": 2.3494,
      "step": 9680
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0001182096803134798,
      "loss": 2.3407,
      "step": 9690
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00011806738009912941,
      "loss": 2.3175,
      "step": 9700
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00011792504205487875,
      "loss": 2.357,
      "step": 9710
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00011778266647875853,
      "loss": 2.3244,
      "step": 9720
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00011764025366887808,
      "loss": 2.358,
      "step": 9730
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00011749780392342459,
      "loss": 2.3311,
      "step": 9740
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00011735531754066271,
      "loss": 2.3168,
      "step": 9750
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00011721279481893368,
      "loss": 2.3429,
      "step": 9760
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0001170702360566549,
      "loss": 2.3225,
      "step": 9770
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00011692764155231926,
      "loss": 2.3359,
      "step": 9780
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0001167850116044944,
      "loss": 2.3153,
      "step": 9790
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00011664234651182222,
      "loss": 2.3389,
      "step": 9800
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00011649964657301824,
      "loss": 2.3372,
      "step": 9810
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00011635691208687092,
      "loss": 2.3199,
      "step": 9820
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00011621414335224096,
      "loss": 2.3124,
      "step": 9830
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00011607134066806093,
      "loss": 2.3238,
      "step": 9840
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00011592850433333438,
      "loss": 2.3227,
      "step": 9850
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00011578563464713537,
      "loss": 2.3387,
      "step": 9860
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00011564273190860772,
      "loss": 2.3425,
      "step": 9870
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00011549979641696456,
      "loss": 2.3328,
      "step": 9880
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0001153568284714875,
      "loss": 2.3353,
      "step": 9890
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00011521382837152614,
      "loss": 2.3266,
      "step": 9900
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0001150707964164975,
      "loss": 2.339,
      "step": 9910
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0001149277329058851,
      "loss": 2.3144,
      "step": 9920
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0001147846381392387,
      "loss": 2.3437,
      "step": 9930
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00011464151241617345,
      "loss": 2.3251,
      "step": 9940
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00011449835603636926,
      "loss": 2.3323,
      "step": 9950
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00011435516929957035,
      "loss": 2.3331,
      "step": 9960
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00011421195250558442,
      "loss": 2.3385,
      "step": 9970
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0001140687059542821,
      "loss": 2.3485,
      "step": 9980
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00011392542994559638,
      "loss": 2.3415,
      "step": 9990
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00011378212477952188,
      "loss": 2.3547,
      "step": 10000
    },
    {
      "epoch": 0.46,
      "eval_accuracy": 0.5437904654599489,
      "eval_loss": 2.188946008682251,
      "eval_runtime": 10.6182,
      "eval_samples_per_second": 124.033,
      "eval_steps_per_second": 1.036,
      "step": 10000
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00011363879075611426,
      "loss": 2.3236,
      "step": 10010
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00011349542817548966,
      "loss": 2.3131,
      "step": 10020
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00011335203733782396,
      "loss": 2.3228,
      "step": 10030
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00011320861854335226,
      "loss": 2.3258,
      "step": 10040
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00011306517209236814,
      "loss": 2.3135,
      "step": 10050
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00011292169828522313,
      "loss": 2.3359,
      "step": 10060
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.000112778197422326,
      "loss": 2.3428,
      "step": 10070
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00011263466980414221,
      "loss": 2.3336,
      "step": 10080
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0001124911157311932,
      "loss": 2.3513,
      "step": 10090
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00011234753550405584,
      "loss": 2.3385,
      "step": 10100
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00011220392942336173,
      "loss": 2.3467,
      "step": 10110
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00011206029778979663,
      "loss": 2.3522,
      "step": 10120
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00011191664090409979,
      "loss": 2.3512,
      "step": 10130
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00011177295906706336,
      "loss": 2.3318,
      "step": 10140
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00011162925257953166,
      "loss": 2.3364,
      "step": 10150
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00011148552174240073,
      "loss": 2.3263,
      "step": 10160
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00011134176685661748,
      "loss": 2.3265,
      "step": 10170
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00011119798822317924,
      "loss": 2.3343,
      "step": 10180
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0001110541861431331,
      "loss": 2.3139,
      "step": 10190
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00011091036091757514,
      "loss": 2.3179,
      "step": 10200
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00011076651284764998,
      "loss": 2.3462,
      "step": 10210
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00011062264223455007,
      "loss": 2.3327,
      "step": 10220
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00011047874937951506,
      "loss": 2.328,
      "step": 10230
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00011033483458383107,
      "loss": 2.3367,
      "step": 10240
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00011019089814883032,
      "loss": 2.3379,
      "step": 10250
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00011004694037589023,
      "loss": 2.287,
      "step": 10260
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00010990296156643294,
      "loss": 2.3171,
      "step": 10270
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00010975896202192462,
      "loss": 2.3391,
      "step": 10280
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00010961494204387486,
      "loss": 2.3295,
      "step": 10290
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00010947090193383604,
      "loss": 2.3165,
      "step": 10300
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00010932684199340268,
      "loss": 2.3188,
      "step": 10310
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00010918276252421087,
      "loss": 2.3403,
      "step": 10320
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0001090386638279375,
      "loss": 2.3262,
      "step": 10330
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0001088945462062998,
      "loss": 2.3395,
      "step": 10340
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00010875040996105459,
      "loss": 2.3411,
      "step": 10350
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00010860625539399767,
      "loss": 2.3279,
      "step": 10360
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00010846208280696324,
      "loss": 2.3217,
      "step": 10370
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0001083178925018232,
      "loss": 2.3251,
      "step": 10380
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00010817368478048657,
      "loss": 2.3115,
      "step": 10390
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00010802945994489886,
      "loss": 2.3251,
      "step": 10400
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00010788521829704132,
      "loss": 2.3286,
      "step": 10410
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00010774096013893049,
      "loss": 2.3224,
      "step": 10420
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00010759668577261746,
      "loss": 2.3305,
      "step": 10430
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00010745239550018725,
      "loss": 2.3222,
      "step": 10440
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00010730808962375817,
      "loss": 2.3178,
      "step": 10450
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00010716376844548126,
      "loss": 2.3329,
      "step": 10460
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00010701943226753953,
      "loss": 2.3072,
      "step": 10470
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00010687508139214739,
      "loss": 2.3339,
      "step": 10480
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0001067307161215501,
      "loss": 2.3264,
      "step": 10490
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00010658633675802301,
      "loss": 2.3086,
      "step": 10500
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00010644194360387096,
      "loss": 2.3008,
      "step": 10510
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0001062975369614277,
      "loss": 2.3459,
      "step": 10520
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00010615311713305525,
      "loss": 2.3223,
      "step": 10530
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00010600868442114315,
      "loss": 2.321,
      "step": 10540
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00010586423912810801,
      "loss": 2.329,
      "step": 10550
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00010571978155639273,
      "loss": 2.3204,
      "step": 10560
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00010557531200846596,
      "loss": 2.3127,
      "step": 10570
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0001054308307868213,
      "loss": 2.3208,
      "step": 10580
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00010528633819397697,
      "loss": 2.3143,
      "step": 10590
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0001051418345324749,
      "loss": 2.3253,
      "step": 10600
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00010499732010488024,
      "loss": 2.3292,
      "step": 10610
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0001048527952137806,
      "loss": 2.3198,
      "step": 10620
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00010470826016178559,
      "loss": 2.3259,
      "step": 10630
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00010456371525152607,
      "loss": 2.33,
      "step": 10640
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0001044191607856535,
      "loss": 2.3291,
      "step": 10650
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0001042745970668394,
      "loss": 2.3436,
      "step": 10660
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00010413002439777466,
      "loss": 2.32,
      "step": 10670
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00010398544308116884,
      "loss": 2.3209,
      "step": 10680
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0001038408534197497,
      "loss": 2.3243,
      "step": 10690
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00010369625571626242,
      "loss": 2.3334,
      "step": 10700
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00010355165027346905,
      "loss": 2.3407,
      "step": 10710
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0001034070373941478,
      "loss": 2.3382,
      "step": 10720
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00010326241738109253,
      "loss": 2.3267,
      "step": 10730
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00010311779053711192,
      "loss": 2.3239,
      "step": 10740
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00010297315716502902,
      "loss": 2.2937,
      "step": 10750
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00010282851756768059,
      "loss": 2.3524,
      "step": 10760
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00010268387204791635,
      "loss": 2.3297,
      "step": 10770
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00010253922090859845,
      "loss": 2.3087,
      "step": 10780
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0001023945644526008,
      "loss": 2.3399,
      "step": 10790
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00010224990298280846,
      "loss": 2.3008,
      "step": 10800
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.000102105236802117,
      "loss": 2.3369,
      "step": 10810
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00010196056621343182,
      "loss": 2.3362,
      "step": 10820
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00010181589151966751,
      "loss": 2.3117,
      "step": 10830
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00010167121302374736,
      "loss": 2.3117,
      "step": 10840
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00010152653102860255,
      "loss": 2.3073,
      "step": 10850
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00010138184583717158,
      "loss": 2.323,
      "step": 10860
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00010123715775239975,
      "loss": 2.3223,
      "step": 10870
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0001010924670772382,
      "loss": 2.3245,
      "step": 10880
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00010094777411464373,
      "loss": 2.306,
      "step": 10890
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00010080307916757782,
      "loss": 2.3155,
      "step": 10900
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00010065838253900608,
      "loss": 2.3591,
      "step": 10910
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00010051368453189767,
      "loss": 2.3251,
      "step": 10920
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00010036898544922463,
      "loss": 2.3192,
      "step": 10930
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0001002242855939613,
      "loss": 2.3133,
      "step": 10940
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00010007958526908359,
      "loss": 2.3151,
      "step": 10950
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.993488477756839e-05,
      "loss": 2.3222,
      "step": 10960
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.979018442239296e-05,
      "loss": 2.3251,
      "step": 10970
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.964548450653426e-05,
      "loss": 2.3065,
      "step": 10980
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.950078533296835e-05,
      "loss": 2.3386,
      "step": 10990
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.935608720466966e-05,
      "loss": 2.3296,
      "step": 11000
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.92113904246106e-05,
      "loss": 2.361,
      "step": 11010
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.906669529576055e-05,
      "loss": 2.3285,
      "step": 11020
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.892200212108561e-05,
      "loss": 2.3269,
      "step": 11030
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.877731120354764e-05,
      "loss": 2.3327,
      "step": 11040
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.863262284610392e-05,
      "loss": 2.3421,
      "step": 11050
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.848793735170624e-05,
      "loss": 2.3239,
      "step": 11060
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.834325502330048e-05,
      "loss": 2.3325,
      "step": 11070
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.819857616382581e-05,
      "loss": 2.326,
      "step": 11080
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.805390107621426e-05,
      "loss": 2.3081,
      "step": 11090
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.790923006338982e-05,
      "loss": 2.3398,
      "step": 11100
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.776456342826809e-05,
      "loss": 2.3334,
      "step": 11110
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.761990147375536e-05,
      "loss": 2.3268,
      "step": 11120
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.747524450274826e-05,
      "loss": 2.3239,
      "step": 11130
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.733059281813289e-05,
      "loss": 2.2933,
      "step": 11140
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.71859467227843e-05,
      "loss": 2.3225,
      "step": 11150
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.70413065195658e-05,
      "loss": 2.3225,
      "step": 11160
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.68966725113285e-05,
      "loss": 2.3043,
      "step": 11170
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.675204500091041e-05,
      "loss": 2.3365,
      "step": 11180
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.660742429113592e-05,
      "loss": 2.3251,
      "step": 11190
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.646281068481532e-05,
      "loss": 2.3294,
      "step": 11200
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.631820448474386e-05,
      "loss": 2.3111,
      "step": 11210
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.617360599370143e-05,
      "loss": 2.3297,
      "step": 11220
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.602901551445167e-05,
      "loss": 2.3262,
      "step": 11230
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.588443334974151e-05,
      "loss": 2.3244,
      "step": 11240
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.573985980230037e-05,
      "loss": 2.3242,
      "step": 11250
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.559529517483983e-05,
      "loss": 2.3113,
      "step": 11260
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.545073977005254e-05,
      "loss": 2.3404,
      "step": 11270
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.530619389061207e-05,
      "loss": 2.3105,
      "step": 11280
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.516165783917187e-05,
      "loss": 2.3289,
      "step": 11290
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.50171319183649e-05,
      "loss": 2.3197,
      "step": 11300
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.487261643080292e-05,
      "loss": 2.3298,
      "step": 11310
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.472811167907581e-05,
      "loss": 2.3388,
      "step": 11320
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.45836179657509e-05,
      "loss": 2.3218,
      "step": 11330
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.443913559337261e-05,
      "loss": 2.3367,
      "step": 11340
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.429466486446137e-05,
      "loss": 2.323,
      "step": 11350
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.415020608151345e-05,
      "loss": 2.3319,
      "step": 11360
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.400575954699995e-05,
      "loss": 2.3079,
      "step": 11370
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.386132556336635e-05,
      "loss": 2.2967,
      "step": 11380
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.371690443303198e-05,
      "loss": 2.3178,
      "step": 11390
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.357249645838905e-05,
      "loss": 2.3233,
      "step": 11400
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.342810194180241e-05,
      "loss": 2.3219,
      "step": 11410
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.32837211856086e-05,
      "loss": 2.3279,
      "step": 11420
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.313935449211546e-05,
      "loss": 2.3196,
      "step": 11430
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.299500216360125e-05,
      "loss": 2.3096,
      "step": 11440
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.28506645023143e-05,
      "loss": 2.3127,
      "step": 11450
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.270634181047214e-05,
      "loss": 2.3342,
      "step": 11460
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.256203439026096e-05,
      "loss": 2.3316,
      "step": 11470
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.241774254383498e-05,
      "loss": 2.3219,
      "step": 11480
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.227346657331589e-05,
      "loss": 2.3233,
      "step": 11490
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.212920678079195e-05,
      "loss": 2.3039,
      "step": 11500
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.198496346831779e-05,
      "loss": 2.313,
      "step": 11510
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.184073693791331e-05,
      "loss": 2.3463,
      "step": 11520
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.169652749156345e-05,
      "loss": 2.3184,
      "step": 11530
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.155233543121725e-05,
      "loss": 2.324,
      "step": 11540
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.140816105878742e-05,
      "loss": 2.3043,
      "step": 11550
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.126400467614956e-05,
      "loss": 2.338,
      "step": 11560
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.111986658514175e-05,
      "loss": 2.3335,
      "step": 11570
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.09757470875636e-05,
      "loss": 2.3216,
      "step": 11580
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.083164648517584e-05,
      "loss": 2.3223,
      "step": 11590
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.068756507969974e-05,
      "loss": 2.3095,
      "step": 11600
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.054350317281623e-05,
      "loss": 2.326,
      "step": 11610
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.039946106616553e-05,
      "loss": 2.3118,
      "step": 11620
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.02554390613463e-05,
      "loss": 2.3286,
      "step": 11630
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.011143745991522e-05,
      "loss": 2.3335,
      "step": 11640
    },
    {
      "epoch": 0.53,
      "learning_rate": 8.99674565633861e-05,
      "loss": 2.3235,
      "step": 11650
    },
    {
      "epoch": 0.53,
      "learning_rate": 8.982349667322957e-05,
      "loss": 2.3166,
      "step": 11660
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.967955809087213e-05,
      "loss": 2.3259,
      "step": 11670
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.95356411176958e-05,
      "loss": 2.3218,
      "step": 11680
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.939174605503719e-05,
      "loss": 2.3175,
      "step": 11690
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.924787320418724e-05,
      "loss": 2.3011,
      "step": 11700
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.910402286639017e-05,
      "loss": 2.3041,
      "step": 11710
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.896019534284321e-05,
      "loss": 2.3196,
      "step": 11720
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.881639093469572e-05,
      "loss": 2.3146,
      "step": 11730
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.867260994304881e-05,
      "loss": 2.3137,
      "step": 11740
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.852885266895433e-05,
      "loss": 2.3096,
      "step": 11750
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.838511941341472e-05,
      "loss": 2.3291,
      "step": 11760
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.82414104773819e-05,
      "loss": 2.3255,
      "step": 11770
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.80977261617571e-05,
      "loss": 2.3145,
      "step": 11780
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.795406676738983e-05,
      "loss": 2.3358,
      "step": 11790
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.78104325950774e-05,
      "loss": 2.312,
      "step": 11800
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.766682394556447e-05,
      "loss": 2.3181,
      "step": 11810
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.75232411195421e-05,
      "loss": 2.325,
      "step": 11820
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.737968441764738e-05,
      "loss": 2.3073,
      "step": 11830
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.723615414046267e-05,
      "loss": 2.3108,
      "step": 11840
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.709265058851501e-05,
      "loss": 2.2938,
      "step": 11850
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.694917406227541e-05,
      "loss": 2.3231,
      "step": 11860
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.680572486215845e-05,
      "loss": 2.3185,
      "step": 11870
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.666230328852134e-05,
      "loss": 2.3233,
      "step": 11880
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.651890964166352e-05,
      "loss": 2.3289,
      "step": 11890
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.637554422182589e-05,
      "loss": 2.3115,
      "step": 11900
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.623220732919038e-05,
      "loss": 2.3082,
      "step": 11910
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.608889926387903e-05,
      "loss": 2.3248,
      "step": 11920
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.594562032595365e-05,
      "loss": 2.325,
      "step": 11930
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.580237081541497e-05,
      "loss": 2.3202,
      "step": 11940
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.56591510322022e-05,
      "loss": 2.3097,
      "step": 11950
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.55159612761922e-05,
      "loss": 2.3318,
      "step": 11960
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.537280184719904e-05,
      "loss": 2.3225,
      "step": 11970
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.522967304497325e-05,
      "loss": 2.3159,
      "step": 11980
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.508657516920117e-05,
      "loss": 2.3115,
      "step": 11990
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.494350851950458e-05,
      "loss": 2.3271,
      "step": 12000
    },
    {
      "epoch": 0.55,
      "eval_accuracy": 0.5458657409572245,
      "eval_loss": 2.175852060317993,
      "eval_runtime": 10.6378,
      "eval_samples_per_second": 123.804,
      "eval_steps_per_second": 1.034,
      "step": 12000
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.480047339543964e-05,
      "loss": 2.3019,
      "step": 12010
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.465747009649674e-05,
      "loss": 2.3274,
      "step": 12020
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.451449892209938e-05,
      "loss": 2.2891,
      "step": 12030
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.437156017160403e-05,
      "loss": 2.3138,
      "step": 12040
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.422865414429912e-05,
      "loss": 2.2956,
      "step": 12050
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.408578113940463e-05,
      "loss": 2.3327,
      "step": 12060
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.394294145607131e-05,
      "loss": 2.304,
      "step": 12070
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.380013539338032e-05,
      "loss": 2.2892,
      "step": 12080
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.365736325034222e-05,
      "loss": 2.3222,
      "step": 12090
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.35146253258967e-05,
      "loss": 2.3145,
      "step": 12100
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.33719219189117e-05,
      "loss": 2.3297,
      "step": 12110
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.322925332818295e-05,
      "loss": 2.325,
      "step": 12120
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.308661985243324e-05,
      "loss": 2.2909,
      "step": 12130
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.294402179031192e-05,
      "loss": 2.3282,
      "step": 12140
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.280145944039401e-05,
      "loss": 2.2954,
      "step": 12150
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.265893310118e-05,
      "loss": 2.3196,
      "step": 12160
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.251644307109474e-05,
      "loss": 2.296,
      "step": 12170
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.237398964848726e-05,
      "loss": 2.3298,
      "step": 12180
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.223157313162981e-05,
      "loss": 2.3163,
      "step": 12190
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.208919381871735e-05,
      "loss": 2.316,
      "step": 12200
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.194685200786707e-05,
      "loss": 2.3233,
      "step": 12210
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.180454799711751e-05,
      "loss": 2.3365,
      "step": 12220
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.166228208442817e-05,
      "loss": 2.3212,
      "step": 12230
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.152005456767861e-05,
      "loss": 2.3161,
      "step": 12240
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.137786574466826e-05,
      "loss": 2.304,
      "step": 12250
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.123571591311525e-05,
      "loss": 2.2967,
      "step": 12260
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.109360537065632e-05,
      "loss": 2.323,
      "step": 12270
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.095153441484571e-05,
      "loss": 2.3459,
      "step": 12280
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.080950334315496e-05,
      "loss": 2.3108,
      "step": 12290
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.0667512452972e-05,
      "loss": 2.3404,
      "step": 12300
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.05255620416007e-05,
      "loss": 2.3181,
      "step": 12310
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.038365240626004e-05,
      "loss": 2.2944,
      "step": 12320
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.024178384408382e-05,
      "loss": 2.3162,
      "step": 12330
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.009995665211963e-05,
      "loss": 2.2941,
      "step": 12340
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.995817112732861e-05,
      "loss": 2.2952,
      "step": 12350
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.981642756658458e-05,
      "loss": 2.3299,
      "step": 12360
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.967472626667348e-05,
      "loss": 2.3268,
      "step": 12370
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.953306752429278e-05,
      "loss": 2.2929,
      "step": 12380
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.939145163605085e-05,
      "loss": 2.307,
      "step": 12390
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.924987889846636e-05,
      "loss": 2.297,
      "step": 12400
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.910834960796752e-05,
      "loss": 2.2928,
      "step": 12410
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.896686406089175e-05,
      "loss": 2.3357,
      "step": 12420
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.882542255348468e-05,
      "loss": 2.3131,
      "step": 12430
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.868402538189989e-05,
      "loss": 2.3266,
      "step": 12440
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.854267284219805e-05,
      "loss": 2.3137,
      "step": 12450
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.84013652303464e-05,
      "loss": 2.3135,
      "step": 12460
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.826010284221802e-05,
      "loss": 2.3032,
      "step": 12470
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.811888597359149e-05,
      "loss": 2.3219,
      "step": 12480
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.79777149201499e-05,
      "loss": 2.3129,
      "step": 12490
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.783658997748054e-05,
      "loss": 2.3234,
      "step": 12500
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.769551144107403e-05,
      "loss": 2.3221,
      "step": 12510
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.755447960632394e-05,
      "loss": 2.3042,
      "step": 12520
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.741349476852598e-05,
      "loss": 2.3078,
      "step": 12530
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.72725572228775e-05,
      "loss": 2.3131,
      "step": 12540
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.713166726447675e-05,
      "loss": 2.3146,
      "step": 12550
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.699082518832246e-05,
      "loss": 2.3017,
      "step": 12560
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.685003128931299e-05,
      "loss": 2.3062,
      "step": 12570
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.670928586224594e-05,
      "loss": 2.3198,
      "step": 12580
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.656858920181734e-05,
      "loss": 2.3028,
      "step": 12590
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.642794160262106e-05,
      "loss": 2.3043,
      "step": 12600
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.628734335914839e-05,
      "loss": 2.3187,
      "step": 12610
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.614679476578716e-05,
      "loss": 2.311,
      "step": 12620
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.600629611682131e-05,
      "loss": 2.3086,
      "step": 12630
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.58658477064301e-05,
      "loss": 2.3117,
      "step": 12640
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.572544982868778e-05,
      "loss": 2.2931,
      "step": 12650
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.558510277756259e-05,
      "loss": 2.3126,
      "step": 12660
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.544480684691654e-05,
      "loss": 2.3062,
      "step": 12670
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.530456233050442e-05,
      "loss": 2.3177,
      "step": 12680
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.516436952197355e-05,
      "loss": 2.3034,
      "step": 12690
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.502422871486282e-05,
      "loss": 2.3085,
      "step": 12700
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.488414020260235e-05,
      "loss": 2.2849,
      "step": 12710
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.474410427851267e-05,
      "loss": 2.3199,
      "step": 12720
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.46041212358043e-05,
      "loss": 2.2908,
      "step": 12730
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.446419136757695e-05,
      "loss": 2.3071,
      "step": 12740
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.432431496681908e-05,
      "loss": 2.3305,
      "step": 12750
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.418449232640708e-05,
      "loss": 2.3035,
      "step": 12760
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.404472373910493e-05,
      "loss": 2.318,
      "step": 12770
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.390500949756329e-05,
      "loss": 2.3143,
      "step": 12780
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.376534989431907e-05,
      "loss": 2.3345,
      "step": 12790
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.362574522179484e-05,
      "loss": 2.3115,
      "step": 12800
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.348619577229804e-05,
      "loss": 2.3191,
      "step": 12810
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.334670183802064e-05,
      "loss": 2.2909,
      "step": 12820
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.320726371103817e-05,
      "loss": 2.3126,
      "step": 12830
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.306788168330954e-05,
      "loss": 2.3198,
      "step": 12840
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.292855604667599e-05,
      "loss": 2.3077,
      "step": 12850
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.278928709286082e-05,
      "loss": 2.3028,
      "step": 12860
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.265007511346859e-05,
      "loss": 2.3104,
      "step": 12870
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.251092039998459e-05,
      "loss": 2.3405,
      "step": 12880
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.237182324377412e-05,
      "loss": 2.3247,
      "step": 12890
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.223278393608216e-05,
      "loss": 2.3115,
      "step": 12900
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.209380276803232e-05,
      "loss": 2.3153,
      "step": 12910
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.195488003062671e-05,
      "loss": 2.3048,
      "step": 12920
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.18160160147449e-05,
      "loss": 2.3083,
      "step": 12930
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.167721101114364e-05,
      "loss": 2.2995,
      "step": 12940
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.153846531045605e-05,
      "loss": 2.3199,
      "step": 12950
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.139977920319112e-05,
      "loss": 2.2871,
      "step": 12960
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.1261152979733e-05,
      "loss": 2.3065,
      "step": 12970
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.112258693034058e-05,
      "loss": 2.3207,
      "step": 12980
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.09840813451466e-05,
      "loss": 2.3323,
      "step": 12990
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.084563651415728e-05,
      "loss": 2.311,
      "step": 13000
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.070725272725167e-05,
      "loss": 2.304,
      "step": 13010
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.056893027418092e-05,
      "loss": 2.3209,
      "step": 13020
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.04306694445678e-05,
      "loss": 2.277,
      "step": 13030
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.029247052790605e-05,
      "loss": 2.3236,
      "step": 13040
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.015433381355981e-05,
      "loss": 2.311,
      "step": 13050
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.001625959076284e-05,
      "loss": 2.3165,
      "step": 13060
    },
    {
      "epoch": 0.6,
      "learning_rate": 6.987824814861828e-05,
      "loss": 2.2844,
      "step": 13070
    },
    {
      "epoch": 0.6,
      "learning_rate": 6.97402997760976e-05,
      "loss": 2.3302,
      "step": 13080
    },
    {
      "epoch": 0.6,
      "learning_rate": 6.960241476204039e-05,
      "loss": 2.2867,
      "step": 13090
    },
    {
      "epoch": 0.6,
      "learning_rate": 6.946459339515343e-05,
      "loss": 2.3083,
      "step": 13100
    },
    {
      "epoch": 0.6,
      "learning_rate": 6.932683596401032e-05,
      "loss": 2.3313,
      "step": 13110
    },
    {
      "epoch": 0.6,
      "learning_rate": 6.918914275705077e-05,
      "loss": 2.3127,
      "step": 13120
    },
    {
      "epoch": 0.6,
      "learning_rate": 6.905151406258003e-05,
      "loss": 2.2841,
      "step": 13130
    },
    {
      "epoch": 0.6,
      "learning_rate": 6.891395016876823e-05,
      "loss": 2.2911,
      "step": 13140
    },
    {
      "epoch": 0.6,
      "learning_rate": 6.87764513636499e-05,
      "loss": 2.3052,
      "step": 13150
    },
    {
      "epoch": 0.6,
      "learning_rate": 6.863901793512315e-05,
      "loss": 2.306,
      "step": 13160
    },
    {
      "epoch": 0.6,
      "learning_rate": 6.850165017094936e-05,
      "loss": 2.2995,
      "step": 13170
    },
    {
      "epoch": 0.6,
      "learning_rate": 6.836434835875234e-05,
      "loss": 2.3245,
      "step": 13180
    },
    {
      "epoch": 0.6,
      "learning_rate": 6.822711278601777e-05,
      "loss": 2.3026,
      "step": 13190
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.80899437400927e-05,
      "loss": 2.3274,
      "step": 13200
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.795284150818489e-05,
      "loss": 2.3085,
      "step": 13210
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.781580637736216e-05,
      "loss": 2.3136,
      "step": 13220
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.767883863455182e-05,
      "loss": 2.3212,
      "step": 13230
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.754193856654016e-05,
      "loss": 2.2962,
      "step": 13240
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.74051064599717e-05,
      "loss": 2.3109,
      "step": 13250
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.72683426013487e-05,
      "loss": 2.2984,
      "step": 13260
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.71316472770305e-05,
      "loss": 2.3213,
      "step": 13270
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.699502077323297e-05,
      "loss": 2.3236,
      "step": 13280
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.685846337602776e-05,
      "loss": 2.3148,
      "step": 13290
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.672197537134204e-05,
      "loss": 2.2987,
      "step": 13300
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.658555704495748e-05,
      "loss": 2.3213,
      "step": 13310
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.644920868251003e-05,
      "loss": 2.3146,
      "step": 13320
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.631293056948895e-05,
      "loss": 2.3265,
      "step": 13330
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.617672299123663e-05,
      "loss": 2.3107,
      "step": 13340
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.604058623294757e-05,
      "loss": 2.3034,
      "step": 13350
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.590452057966816e-05,
      "loss": 2.3502,
      "step": 13360
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.576852631629573e-05,
      "loss": 2.2922,
      "step": 13370
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.563260372757833e-05,
      "loss": 2.3409,
      "step": 13380
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.549675309811379e-05,
      "loss": 2.3116,
      "step": 13390
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.536097471234927e-05,
      "loss": 2.3259,
      "step": 13400
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.522526885458078e-05,
      "loss": 2.3194,
      "step": 13410
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.508963580895233e-05,
      "loss": 2.3174,
      "step": 13420
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.495407585945563e-05,
      "loss": 2.3057,
      "step": 13430
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.481858928992918e-05,
      "loss": 2.321,
      "step": 13440
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.468317638405794e-05,
      "loss": 2.3236,
      "step": 13450
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.454783742537255e-05,
      "loss": 2.2912,
      "step": 13460
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.441257269724896e-05,
      "loss": 2.324,
      "step": 13470
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.427738248290748e-05,
      "loss": 2.3173,
      "step": 13480
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.414226706541267e-05,
      "loss": 2.3069,
      "step": 13490
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.400722672767216e-05,
      "loss": 2.3107,
      "step": 13500
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.387226175243672e-05,
      "loss": 2.2888,
      "step": 13510
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.373737242229903e-05,
      "loss": 2.2753,
      "step": 13520
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.360255901969354e-05,
      "loss": 2.308,
      "step": 13530
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.346782182689566e-05,
      "loss": 2.3152,
      "step": 13540
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.333316112602132e-05,
      "loss": 2.319,
      "step": 13550
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.319857719902614e-05,
      "loss": 2.3116,
      "step": 13560
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.306407032770516e-05,
      "loss": 2.296,
      "step": 13570
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.292964079369192e-05,
      "loss": 2.3127,
      "step": 13580
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.279528887845815e-05,
      "loss": 2.304,
      "step": 13590
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.266101486331302e-05,
      "loss": 2.3073,
      "step": 13600
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.252681902940252e-05,
      "loss": 2.3044,
      "step": 13610
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.239270165770908e-05,
      "loss": 2.3053,
      "step": 13620
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.225866302905066e-05,
      "loss": 2.3229,
      "step": 13630
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.212470342408058e-05,
      "loss": 2.3171,
      "step": 13640
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.19908231232865e-05,
      "loss": 2.3066,
      "step": 13650
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.185702240699014e-05,
      "loss": 2.3237,
      "step": 13660
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.172330155534651e-05,
      "loss": 2.2863,
      "step": 13670
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.158966084834348e-05,
      "loss": 2.308,
      "step": 13680
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.145610056580105e-05,
      "loss": 2.3156,
      "step": 13690
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.132262098737083e-05,
      "loss": 2.3101,
      "step": 13700
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.118922239253544e-05,
      "loss": 2.3016,
      "step": 13710
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.105590506060806e-05,
      "loss": 2.317,
      "step": 13720
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.092266927073147e-05,
      "loss": 2.3091,
      "step": 13730
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.0789515301877995e-05,
      "loss": 2.3037,
      "step": 13740
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.065644343284844e-05,
      "loss": 2.2933,
      "step": 13750
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.052345394227178e-05,
      "loss": 2.3054,
      "step": 13760
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.039054710860451e-05,
      "loss": 2.32,
      "step": 13770
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.025772321013008e-05,
      "loss": 2.3083,
      "step": 13780
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.0124982524958205e-05,
      "loss": 2.321,
      "step": 13790
    },
    {
      "epoch": 0.63,
      "learning_rate": 5.999232533102439e-05,
      "loss": 2.3379,
      "step": 13800
    },
    {
      "epoch": 0.63,
      "learning_rate": 5.985975190608946e-05,
      "loss": 2.3056,
      "step": 13810
    },
    {
      "epoch": 0.63,
      "learning_rate": 5.9727262527738595e-05,
      "loss": 2.2938,
      "step": 13820
    },
    {
      "epoch": 0.63,
      "learning_rate": 5.959485747338126e-05,
      "loss": 2.2844,
      "step": 13830
    },
    {
      "epoch": 0.63,
      "learning_rate": 5.9462537020250155e-05,
      "loss": 2.2905,
      "step": 13840
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.933030144540094e-05,
      "loss": 2.301,
      "step": 13850
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.9198151025711534e-05,
      "loss": 2.3103,
      "step": 13860
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.90660860378816e-05,
      "loss": 2.2778,
      "step": 13870
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.893410675843176e-05,
      "loss": 2.2887,
      "step": 13880
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.880221346370345e-05,
      "loss": 2.3046,
      "step": 13890
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.867040642985777e-05,
      "loss": 2.3042,
      "step": 13900
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.853868593287548e-05,
      "loss": 2.3081,
      "step": 13910
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.840705224855592e-05,
      "loss": 2.3084,
      "step": 13920
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.8275505652516806e-05,
      "loss": 2.3106,
      "step": 13930
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.814404642019341e-05,
      "loss": 2.3253,
      "step": 13940
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.801267482683821e-05,
      "loss": 2.3085,
      "step": 13950
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.788139114752005e-05,
      "loss": 2.3029,
      "step": 13960
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.7750195657123743e-05,
      "loss": 2.3,
      "step": 13970
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.7619088630349416e-05,
      "loss": 2.3055,
      "step": 13980
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.7488070341712106e-05,
      "loss": 2.3151,
      "step": 13990
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.73571410655409e-05,
      "loss": 2.3153,
      "step": 14000
    },
    {
      "epoch": 0.64,
      "eval_accuracy": 0.547557283467343,
      "eval_loss": 2.1664211750030518,
      "eval_runtime": 10.6183,
      "eval_samples_per_second": 124.031,
      "eval_steps_per_second": 1.036,
      "step": 14000
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.7226301075978504e-05,
      "loss": 2.2999,
      "step": 14010
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.7095550646980834e-05,
      "loss": 2.3314,
      "step": 14020
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.696489005231607e-05,
      "loss": 2.3378,
      "step": 14030
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.6834319565564506e-05,
      "loss": 2.3276,
      "step": 14040
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.670383946011757e-05,
      "loss": 2.3035,
      "step": 14050
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.6573450009177595e-05,
      "loss": 2.3095,
      "step": 14060
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.6443151485757005e-05,
      "loss": 2.3013,
      "step": 14070
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.631294416267794e-05,
      "loss": 2.3074,
      "step": 14080
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.6182828312571424e-05,
      "loss": 2.3108,
      "step": 14090
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.605280420787715e-05,
      "loss": 2.3163,
      "step": 14100
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.592287212084251e-05,
      "loss": 2.2957,
      "step": 14110
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.579303232352241e-05,
      "loss": 2.307,
      "step": 14120
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.56632850877784e-05,
      "loss": 2.294,
      "step": 14130
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.553363068527826e-05,
      "loss": 2.3065,
      "step": 14140
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.5404069387495336e-05,
      "loss": 2.3169,
      "step": 14150
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.52746014657082e-05,
      "loss": 2.3231,
      "step": 14160
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.514522719099967e-05,
      "loss": 2.3279,
      "step": 14170
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.5015946834256704e-05,
      "loss": 2.3173,
      "step": 14180
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.488676066616949e-05,
      "loss": 2.3125,
      "step": 14190
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.4757668957230977e-05,
      "loss": 2.3106,
      "step": 14200
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.462867197773647e-05,
      "loss": 2.3074,
      "step": 14210
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.4499769997782746e-05,
      "loss": 2.301,
      "step": 14220
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.4370963287267874e-05,
      "loss": 2.2743,
      "step": 14230
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.424225211589024e-05,
      "loss": 2.314,
      "step": 14240
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.411363675314836e-05,
      "loss": 2.2887,
      "step": 14250
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.3985117468339986e-05,
      "loss": 2.3154,
      "step": 14260
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.385669453056189e-05,
      "loss": 2.293,
      "step": 14270
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.37283682087089e-05,
      "loss": 2.3178,
      "step": 14280
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.360013877147374e-05,
      "loss": 2.3154,
      "step": 14290
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.347200648734614e-05,
      "loss": 2.3046,
      "step": 14300
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.3343971624612446e-05,
      "loss": 2.3247,
      "step": 14310
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.3216034451355004e-05,
      "loss": 2.3111,
      "step": 14320
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.30881952354517e-05,
      "loss": 2.319,
      "step": 14330
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.296045424457521e-05,
      "loss": 2.297,
      "step": 14340
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.2832811746192635e-05,
      "loss": 2.3038,
      "step": 14350
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.270526800756475e-05,
      "loss": 2.2795,
      "step": 14360
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.2577823295745646e-05,
      "loss": 2.3192,
      "step": 14370
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.245047787758197e-05,
      "loss": 2.323,
      "step": 14380
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.232323201971257e-05,
      "loss": 2.2988,
      "step": 14390
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.219608598856778e-05,
      "loss": 2.3088,
      "step": 14400
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.206904005036884e-05,
      "loss": 2.2934,
      "step": 14410
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.194209447112759e-05,
      "loss": 2.2837,
      "step": 14420
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.1815249516645536e-05,
      "loss": 2.3019,
      "step": 14430
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.168850545251367e-05,
      "loss": 2.312,
      "step": 14440
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.156186254411159e-05,
      "loss": 2.2882,
      "step": 14450
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.143532105660724e-05,
      "loss": 2.3297,
      "step": 14460
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.13088812549561e-05,
      "loss": 2.3327,
      "step": 14470
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.118254340390072e-05,
      "loss": 2.313,
      "step": 14480
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.1056307767970235e-05,
      "loss": 2.2993,
      "step": 14490
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.09301746114798e-05,
      "loss": 2.3118,
      "step": 14500
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.080414419852988e-05,
      "loss": 2.3066,
      "step": 14510
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.067821679300597e-05,
      "loss": 2.3277,
      "step": 14520
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.055239265857772e-05,
      "loss": 2.3103,
      "step": 14530
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.0426672058698685e-05,
      "loss": 2.3147,
      "step": 14540
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.030105525660554e-05,
      "loss": 2.3035,
      "step": 14550
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.017554251531768e-05,
      "loss": 2.2993,
      "step": 14560
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.005013409763657e-05,
      "loss": 2.2943,
      "step": 14570
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.992483026614533e-05,
      "loss": 2.3149,
      "step": 14580
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.9799631283207995e-05,
      "loss": 2.3263,
      "step": 14590
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.967453741096907e-05,
      "loss": 2.3054,
      "step": 14600
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.954954891135308e-05,
      "loss": 2.2943,
      "step": 14610
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.9424666046063795e-05,
      "loss": 2.2874,
      "step": 14620
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.92998890765839e-05,
      "loss": 2.3065,
      "step": 14630
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.9175218264174315e-05,
      "loss": 2.2987,
      "step": 14640
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.905065386987365e-05,
      "loss": 2.3038,
      "step": 14650
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.8926196154497713e-05,
      "loss": 2.2866,
      "step": 14660
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.8801845378639036e-05,
      "loss": 2.2866,
      "step": 14670
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.8677601802666065e-05,
      "loss": 2.3225,
      "step": 14680
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.8553465686722976e-05,
      "loss": 2.3008,
      "step": 14690
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.8429437290728774e-05,
      "loss": 2.3137,
      "step": 14700
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.8305516874377086e-05,
      "loss": 2.3078,
      "step": 14710
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.8181704697135265e-05,
      "loss": 2.3098,
      "step": 14720
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.8058001018244204e-05,
      "loss": 2.3152,
      "step": 14730
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.793440609671748e-05,
      "loss": 2.2998,
      "step": 14740
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.781092019134107e-05,
      "loss": 2.309,
      "step": 14750
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.768754356067255e-05,
      "loss": 2.2867,
      "step": 14760
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.756427646304088e-05,
      "loss": 2.3157,
      "step": 14770
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.744111915654551e-05,
      "loss": 2.3163,
      "step": 14780
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.731807189905608e-05,
      "loss": 2.2974,
      "step": 14790
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.719513494821176e-05,
      "loss": 2.3136,
      "step": 14800
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.7072308561420866e-05,
      "loss": 2.3072,
      "step": 14810
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.6949592995860106e-05,
      "loss": 2.3148,
      "step": 14820
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.6826988508474134e-05,
      "loss": 2.3122,
      "step": 14830
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.6704495355975164e-05,
      "loss": 2.3239,
      "step": 14840
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.658211379484213e-05,
      "loss": 2.3139,
      "step": 14850
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.645984408132045e-05,
      "loss": 2.3075,
      "step": 14860
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.6337686471421225e-05,
      "loss": 2.2895,
      "step": 14870
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.621564122092097e-05,
      "loss": 2.3127,
      "step": 14880
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.60937085853608e-05,
      "loss": 2.3051,
      "step": 14890
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.597188882004616e-05,
      "loss": 2.3095,
      "step": 14900
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.585018218004603e-05,
      "loss": 2.3097,
      "step": 14910
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.5728588920192696e-05,
      "loss": 2.3249,
      "step": 14920
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.5607109295080865e-05,
      "loss": 2.3014,
      "step": 14930
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.548574355906746e-05,
      "loss": 2.311,
      "step": 14940
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.5364491966270864e-05,
      "loss": 2.2995,
      "step": 14950
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.5243354770570465e-05,
      "loss": 2.2954,
      "step": 14960
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.51223322256061e-05,
      "loss": 2.32,
      "step": 14970
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.5001424584777674e-05,
      "loss": 2.2899,
      "step": 14980
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.488063210124436e-05,
      "loss": 2.3119,
      "step": 14990
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.475995502792424e-05,
      "loss": 2.3186,
      "step": 15000
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.4639393617493855e-05,
      "loss": 2.2999,
      "step": 15010
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.4518948122387394e-05,
      "loss": 2.3013,
      "step": 15020
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.439861879479652e-05,
      "loss": 2.3055,
      "step": 15030
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.4278405886669495e-05,
      "loss": 2.3014,
      "step": 15040
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.415830964971098e-05,
      "loss": 2.2959,
      "step": 15050
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.4038330335381195e-05,
      "loss": 2.3173,
      "step": 15060
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.3918468194895677e-05,
      "loss": 2.2939,
      "step": 15070
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.379872347922449e-05,
      "loss": 2.2893,
      "step": 15080
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.3679096439091996e-05,
      "loss": 2.3129,
      "step": 15090
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.355958732497597e-05,
      "loss": 2.3153,
      "step": 15100
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.3440196387107454e-05,
      "loss": 2.29,
      "step": 15110
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.332092387546994e-05,
      "loss": 2.2898,
      "step": 15120
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.320177003979898e-05,
      "loss": 2.278,
      "step": 15130
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.308273512958161e-05,
      "loss": 2.3059,
      "step": 15140
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.296381939405595e-05,
      "loss": 2.2969,
      "step": 15150
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.284502308221047e-05,
      "loss": 2.2898,
      "step": 15160
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.272634644278372e-05,
      "loss": 2.321,
      "step": 15170
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.260778972426352e-05,
      "loss": 2.308,
      "step": 15180
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.248935317488678e-05,
      "loss": 2.2984,
      "step": 15190
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.237103704263864e-05,
      "loss": 2.3131,
      "step": 15200
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.2252841575252135e-05,
      "loss": 2.3112,
      "step": 15210
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.2134767020207765e-05,
      "loss": 2.3169,
      "step": 15220
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.201681362473269e-05,
      "loss": 2.3182,
      "step": 15230
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.189898163580055e-05,
      "loss": 2.3171,
      "step": 15240
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.178127130013062e-05,
      "loss": 2.3055,
      "step": 15250
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.1663682864187624e-05,
      "loss": 2.3057,
      "step": 15260
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.1546216574180876e-05,
      "loss": 2.3231,
      "step": 15270
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.14288726760641e-05,
      "loss": 2.3223,
      "step": 15280
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.1311651415534644e-05,
      "loss": 2.3112,
      "step": 15290
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.11945530380331e-05,
      "loss": 2.2802,
      "step": 15300
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.107757778874274e-05,
      "loss": 2.3204,
      "step": 15310
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.096072591258913e-05,
      "loss": 2.3151,
      "step": 15320
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.0843997654239374e-05,
      "loss": 2.3136,
      "step": 15330
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.0727393258101876e-05,
      "loss": 2.3046,
      "step": 15340
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.0610912968325565e-05,
      "loss": 2.3124,
      "step": 15350
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.049455702879965e-05,
      "loss": 2.2923,
      "step": 15360
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.0378325683152815e-05,
      "loss": 2.2849,
      "step": 15370
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.0262219174753055e-05,
      "loss": 2.294,
      "step": 15380
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.014623774670677e-05,
      "loss": 2.325,
      "step": 15390
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.0030381641858674e-05,
      "loss": 2.3117,
      "step": 15400
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.99146511027909e-05,
      "loss": 2.3086,
      "step": 15410
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.9799046371822726e-05,
      "loss": 2.302,
      "step": 15420
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.968356769101008e-05,
      "loss": 2.3094,
      "step": 15430
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.9568215302144885e-05,
      "loss": 2.2963,
      "step": 15440
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.945298944675462e-05,
      "loss": 2.3276,
      "step": 15450
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.933789036610192e-05,
      "loss": 2.3057,
      "step": 15460
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.9222918301183885e-05,
      "loss": 2.2968,
      "step": 15470
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.9108073492731654e-05,
      "loss": 2.2894,
      "step": 15480
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.8993356181210025e-05,
      "loss": 2.3068,
      "step": 15490
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.8878766606816705e-05,
      "loss": 2.3008,
      "step": 15500
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.8764305009482084e-05,
      "loss": 2.2847,
      "step": 15510
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.864997162886844e-05,
      "loss": 2.3015,
      "step": 15520
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.853576670436973e-05,
      "loss": 2.3248,
      "step": 15530
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.842169047511079e-05,
      "loss": 2.3129,
      "step": 15540
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.830774317994717e-05,
      "loss": 2.2694,
      "step": 15550
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.819392505746431e-05,
      "loss": 2.2874,
      "step": 15560
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.808023634597728e-05,
      "loss": 2.3114,
      "step": 15570
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.7966677283530083e-05,
      "loss": 2.2998,
      "step": 15580
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.78532481078954e-05,
      "loss": 2.2926,
      "step": 15590
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.773994905657384e-05,
      "loss": 2.309,
      "step": 15600
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.762678036679358e-05,
      "loss": 2.3068,
      "step": 15610
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.751374227550982e-05,
      "loss": 2.3137,
      "step": 15620
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.740083501940441e-05,
      "loss": 2.3001,
      "step": 15630
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.7288058834885165e-05,
      "loss": 2.3113,
      "step": 15640
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.717541395808542e-05,
      "loss": 2.3245,
      "step": 15650
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.706290062486374e-05,
      "loss": 2.3196,
      "step": 15660
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.695051907080306e-05,
      "loss": 2.3048,
      "step": 15670
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.6838269531210565e-05,
      "loss": 2.2982,
      "step": 15680
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.672615224111688e-05,
      "loss": 2.3063,
      "step": 15690
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.661416743527587e-05,
      "loss": 2.3045,
      "step": 15700
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.650231534816384e-05,
      "loss": 2.3311,
      "step": 15710
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.6390596213979366e-05,
      "loss": 2.2962,
      "step": 15720
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.62790102666425e-05,
      "loss": 2.2972,
      "step": 15730
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.6167557739794556e-05,
      "loss": 2.2845,
      "step": 15740
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.6056238866797356e-05,
      "loss": 2.3186,
      "step": 15750
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.5945053880733e-05,
      "loss": 2.297,
      "step": 15760
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.5834003014403197e-05,
      "loss": 2.3066,
      "step": 15770
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.572308650032878e-05,
      "loss": 2.2711,
      "step": 15780
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.5612304570749335e-05,
      "loss": 2.3067,
      "step": 15790
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.5501657457622684e-05,
      "loss": 2.2966,
      "step": 15800
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.5391145392624305e-05,
      "loss": 2.3022,
      "step": 15810
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.5280768607146876e-05,
      "loss": 2.2973,
      "step": 15820
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.517052733229994e-05,
      "loss": 2.2985,
      "step": 15830
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.5060421798909195e-05,
      "loss": 2.3133,
      "step": 15840
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.4950452237516216e-05,
      "loss": 2.3151,
      "step": 15850
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.484061887837777e-05,
      "loss": 2.3002,
      "step": 15860
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.473092195146556e-05,
      "loss": 2.296,
      "step": 15870
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.462136168646547e-05,
      "loss": 2.3132,
      "step": 15880
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.451193831277743e-05,
      "loss": 2.3139,
      "step": 15890
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.4402652059514565e-05,
      "loss": 2.2927,
      "step": 15900
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.4293503155503026e-05,
      "loss": 2.3147,
      "step": 15910
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.4184491829281264e-05,
      "loss": 2.3012,
      "step": 15920
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.40756183090998e-05,
      "loss": 2.3054,
      "step": 15930
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.39668828229205e-05,
      "loss": 2.3129,
      "step": 15940
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.385828559841625e-05,
      "loss": 2.3001,
      "step": 15950
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.374982686297041e-05,
      "loss": 2.3276,
      "step": 15960
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.364150684367647e-05,
      "loss": 2.3241,
      "step": 15970
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.353332576733735e-05,
      "loss": 2.3157,
      "step": 15980
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.342528386046514e-05,
      "loss": 2.3026,
      "step": 15990
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.331738134928049e-05,
      "loss": 2.3175,
      "step": 16000
    },
    {
      "epoch": 0.73,
      "eval_accuracy": 0.548459835328819,
      "eval_loss": 2.161773920059204,
      "eval_runtime": 10.6727,
      "eval_samples_per_second": 123.4,
      "eval_steps_per_second": 1.031,
      "step": 16000
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.320961845971212e-05,
      "loss": 2.2986,
      "step": 16010
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.310199541739656e-05,
      "loss": 2.3007,
      "step": 16020
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.2994512447677374e-05,
      "loss": 2.291,
      "step": 16030
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.288716977560495e-05,
      "loss": 2.3156,
      "step": 16040
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.277996762593582e-05,
      "loss": 2.2971,
      "step": 16050
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.267290622313239e-05,
      "loss": 2.3141,
      "step": 16060
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.2565985791362254e-05,
      "loss": 2.2884,
      "step": 16070
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.2459206554497954e-05,
      "loss": 2.3175,
      "step": 16080
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.2352568736116285e-05,
      "loss": 2.3089,
      "step": 16090
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.2246072559498045e-05,
      "loss": 2.3024,
      "step": 16100
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.213971824762737e-05,
      "loss": 2.2867,
      "step": 16110
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.203350602319138e-05,
      "loss": 2.3223,
      "step": 16120
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.19274361085797e-05,
      "loss": 2.2916,
      "step": 16130
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.182150872588401e-05,
      "loss": 2.2983,
      "step": 16140
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.171572409689747e-05,
      "loss": 2.2905,
      "step": 16150
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.161008244311447e-05,
      "loss": 2.3159,
      "step": 16160
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.1504583985729886e-05,
      "loss": 2.316,
      "step": 16170
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.139922894563891e-05,
      "loss": 2.3099,
      "step": 16180
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.1294017543436304e-05,
      "loss": 2.3322,
      "step": 16190
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.1188949999416203e-05,
      "loss": 2.3297,
      "step": 16200
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.108402653357145e-05,
      "loss": 2.2975,
      "step": 16210
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.097924736559321e-05,
      "loss": 2.2976,
      "step": 16220
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.087461271487059e-05,
      "loss": 2.3075,
      "step": 16230
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.077012280048999e-05,
      "loss": 2.3148,
      "step": 16240
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.066577784123489e-05,
      "loss": 2.2901,
      "step": 16250
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.0561578055585164e-05,
      "loss": 2.3105,
      "step": 16260
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.04575236617167e-05,
      "loss": 2.2884,
      "step": 16270
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.035361487750107e-05,
      "loss": 2.3152,
      "step": 16280
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.024985192050488e-05,
      "loss": 2.291,
      "step": 16290
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.0146235007989376e-05,
      "loss": 2.2914,
      "step": 16300
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.0042764356910137e-05,
      "loss": 2.3059,
      "step": 16310
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.9939440183916334e-05,
      "loss": 2.2772,
      "step": 16320
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.9836262705350594e-05,
      "loss": 2.2922,
      "step": 16330
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.9733232137248258e-05,
      "loss": 2.307,
      "step": 16340
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.9630348695337184e-05,
      "loss": 2.3241,
      "step": 16350
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.9527612595037045e-05,
      "loss": 2.2688,
      "step": 16360
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.9425024051459138e-05,
      "loss": 2.2978,
      "step": 16370
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.932258327940569e-05,
      "loss": 2.2968,
      "step": 16380
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.922029049336963e-05,
      "loss": 2.3018,
      "step": 16390
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.911814590753391e-05,
      "loss": 2.3042,
      "step": 16400
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.901614973577125e-05,
      "loss": 2.3177,
      "step": 16410
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.8914302191643638e-05,
      "loss": 2.2883,
      "step": 16420
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.88126034884018e-05,
      "loss": 2.3142,
      "step": 16430
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.8711053838984815e-05,
      "loss": 2.3003,
      "step": 16440
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.860965345601977e-05,
      "loss": 2.3007,
      "step": 16450
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.85084025518211e-05,
      "loss": 2.3066,
      "step": 16460
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.840730133839028e-05,
      "loss": 2.3003,
      "step": 16470
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.8306350027415428e-05,
      "loss": 2.3037,
      "step": 16480
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.82055488302707e-05,
      "loss": 2.2898,
      "step": 16490
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.810489795801603e-05,
      "loss": 2.3115,
      "step": 16500
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.8004397621396507e-05,
      "loss": 2.2959,
      "step": 16510
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.790404803084211e-05,
      "loss": 2.3228,
      "step": 16520
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.7803849396467107e-05,
      "loss": 2.3131,
      "step": 16530
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.7703801928069772e-05,
      "loss": 2.2947,
      "step": 16540
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.760390583513176e-05,
      "loss": 2.3233,
      "step": 16550
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.75041613268179e-05,
      "loss": 2.3242,
      "step": 16560
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.74045686119755e-05,
      "loss": 2.3177,
      "step": 16570
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.730512789913415e-05,
      "loss": 2.28,
      "step": 16580
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.7205839396505118e-05,
      "loss": 2.3258,
      "step": 16590
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.710670331198094e-05,
      "loss": 2.3015,
      "step": 16600
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.7007719853135082e-05,
      "loss": 2.3047,
      "step": 16610
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.6908889227221378e-05,
      "loss": 2.2958,
      "step": 16620
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.6810211641173742e-05,
      "loss": 2.3081,
      "step": 16630
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.6711687301605537e-05,
      "loss": 2.3224,
      "step": 16640
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.6613316414809377e-05,
      "loss": 2.2959,
      "step": 16650
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.6515099186756443e-05,
      "loss": 2.2976,
      "step": 16660
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.6417035823096315e-05,
      "loss": 2.301,
      "step": 16670
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.631912652915629e-05,
      "loss": 2.3015,
      "step": 16680
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.6221371509941173e-05,
      "loss": 2.3011,
      "step": 16690
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.6123770970132632e-05,
      "loss": 2.2944,
      "step": 16700
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.6026325114089023e-05,
      "loss": 2.2846,
      "step": 16710
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.5929034145844655e-05,
      "loss": 2.2901,
      "step": 16720
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.58318982691097e-05,
      "loss": 2.3167,
      "step": 16730
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.5734917687269455e-05,
      "loss": 2.2903,
      "step": 16740
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.5638092603384156e-05,
      "loss": 2.3095,
      "step": 16750
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.55414232201884e-05,
      "loss": 2.2955,
      "step": 16760
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.5444909740090784e-05,
      "loss": 2.279,
      "step": 16770
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.534855236517344e-05,
      "loss": 2.309,
      "step": 16780
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.5252351297191747e-05,
      "loss": 2.2961,
      "step": 16790
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.5156306737573676e-05,
      "loss": 2.3033,
      "step": 16800
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.50604188874196e-05,
      "loss": 2.312,
      "step": 16810
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.496468794750172e-05,
      "loss": 2.2992,
      "step": 16820
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.486911411826367e-05,
      "loss": 2.3011,
      "step": 16830
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.47736975998202e-05,
      "loss": 2.3055,
      "step": 16840
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.4678438591956576e-05,
      "loss": 2.3156,
      "step": 16850
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.4583337294128393e-05,
      "loss": 2.3154,
      "step": 16860
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.4488393905460892e-05,
      "loss": 2.2837,
      "step": 16870
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.439360862474881e-05,
      "loss": 2.3017,
      "step": 16880
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.429898165045571e-05,
      "loss": 2.282,
      "step": 16890
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.4204513180713796e-05,
      "loss": 2.3139,
      "step": 16900
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.4110203413323306e-05,
      "loss": 2.2838,
      "step": 16910
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.4016052545752267e-05,
      "loss": 2.2969,
      "step": 16920
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.3922060775135923e-05,
      "loss": 2.3137,
      "step": 16930
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.3828228298276435e-05,
      "loss": 2.2933,
      "step": 16940
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.3734555311642393e-05,
      "loss": 2.3056,
      "step": 16950
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.3641042011368518e-05,
      "loss": 2.2857,
      "step": 16960
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.3547688593255102e-05,
      "loss": 2.3257,
      "step": 16970
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.3454495252767727e-05,
      "loss": 2.2987,
      "step": 16980
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.3361462185036742e-05,
      "loss": 2.3047,
      "step": 16990
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.326858958485698e-05,
      "loss": 2.304,
      "step": 17000
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.317587764668725e-05,
      "loss": 2.3215,
      "step": 17010
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.3083326564649897e-05,
      "loss": 2.3158,
      "step": 17020
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.2990936532530616e-05,
      "loss": 2.2929,
      "step": 17030
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.2898707743777738e-05,
      "loss": 2.3008,
      "step": 17040
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.2806640391502086e-05,
      "loss": 2.2874,
      "step": 17050
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.2714734668476377e-05,
      "loss": 2.3024,
      "step": 17060
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.2622990767135012e-05,
      "loss": 2.2978,
      "step": 17070
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.2531408879573457e-05,
      "loss": 2.298,
      "step": 17080
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.2439989197547984e-05,
      "loss": 2.3104,
      "step": 17090
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.2348731912475295e-05,
      "loss": 2.285,
      "step": 17100
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.225763721543198e-05,
      "loss": 2.3173,
      "step": 17110
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.2166705297154234e-05,
      "loss": 2.2826,
      "step": 17120
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.207593634803744e-05,
      "loss": 2.3086,
      "step": 17130
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.1985330558135698e-05,
      "loss": 2.2742,
      "step": 17140
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.189488811716156e-05,
      "loss": 2.2977,
      "step": 17150
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.1804609214485462e-05,
      "loss": 2.3063,
      "step": 17160
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.1714494039135525e-05,
      "loss": 2.3053,
      "step": 17170
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.162454277979694e-05,
      "loss": 2.3253,
      "step": 17180
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.1534755624811796e-05,
      "loss": 2.2944,
      "step": 17190
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.1445132762178477e-05,
      "loss": 2.2964,
      "step": 17200
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.135567437955146e-05,
      "loss": 2.2812,
      "step": 17210
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.126638066424076e-05,
      "loss": 2.3111,
      "step": 17220
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.1177251803211607e-05,
      "loss": 2.2905,
      "step": 17230
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.1088287983084142e-05,
      "loss": 2.2832,
      "step": 17240
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.0999489390132843e-05,
      "loss": 2.3018,
      "step": 17250
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.0910856210286232e-05,
      "loss": 2.3035,
      "step": 17260
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.08223886291266e-05,
      "loss": 2.2931,
      "step": 17270
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.0734086831889377e-05,
      "loss": 2.3126,
      "step": 17280
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.0645951003462906e-05,
      "loss": 2.2971,
      "step": 17290
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.0557981328388063e-05,
      "loss": 2.305,
      "step": 17300
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.0470177990857762e-05,
      "loss": 2.2776,
      "step": 17310
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.038254117471673e-05,
      "loss": 2.3058,
      "step": 17320
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.029507106346089e-05,
      "loss": 2.2888,
      "step": 17330
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.020776784023727e-05,
      "loss": 2.2938,
      "step": 17340
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.0120631687843327e-05,
      "loss": 2.3006,
      "step": 17350
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.0033662788726815e-05,
      "loss": 2.2834,
      "step": 17360
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.9946861324985177e-05,
      "loss": 2.2852,
      "step": 17370
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.98602274783654e-05,
      "loss": 2.2909,
      "step": 17380
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.9773761430263383e-05,
      "loss": 2.3079,
      "step": 17390
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.96874633617238e-05,
      "loss": 2.2947,
      "step": 17400
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.960133345343953e-05,
      "loss": 2.2873,
      "step": 17410
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.9515371885751366e-05,
      "loss": 2.3176,
      "step": 17420
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.9429578838647644e-05,
      "loss": 2.2847,
      "step": 17430
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.9343954491763816e-05,
      "loss": 2.3363,
      "step": 17440
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.925849902438218e-05,
      "loss": 2.3102,
      "step": 17450
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.9173212615431325e-05,
      "loss": 2.2951,
      "step": 17460
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.9088095443485964e-05,
      "loss": 2.3141,
      "step": 17470
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.9003147686766355e-05,
      "loss": 2.3042,
      "step": 17480
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.8918369523138147e-05,
      "loss": 2.3026,
      "step": 17490
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.883376113011177e-05,
      "loss": 2.3113,
      "step": 17500
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.8749322684842306e-05,
      "loss": 2.286,
      "step": 17510
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.8665054364128865e-05,
      "loss": 2.3095,
      "step": 17520
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.858095634441448e-05,
      "loss": 2.2883,
      "step": 17530
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.8497028801785476e-05,
      "loss": 2.2918,
      "step": 17540
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.841327191197135e-05,
      "loss": 2.2789,
      "step": 17550
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.832968585034418e-05,
      "loss": 2.3008,
      "step": 17560
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.824627079191844e-05,
      "loss": 2.2852,
      "step": 17570
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.816302691135051e-05,
      "loss": 2.2731,
      "step": 17580
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.8079954382938334e-05,
      "loss": 2.3033,
      "step": 17590
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7997053380621108e-05,
      "loss": 2.2707,
      "step": 17600
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7914324077978905e-05,
      "loss": 2.2908,
      "step": 17610
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7831766648232252e-05,
      "loss": 2.3059,
      "step": 17620
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.774938126424177e-05,
      "loss": 2.2947,
      "step": 17630
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7667168098507958e-05,
      "loss": 2.2863,
      "step": 17640
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.758512732317059e-05,
      "loss": 2.3019,
      "step": 17650
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7503259110008607e-05,
      "loss": 2.3132,
      "step": 17660
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.742156363043952e-05,
      "loss": 2.3237,
      "step": 17670
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7340041055519275e-05,
      "loss": 2.2949,
      "step": 17680
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7258691555941686e-05,
      "loss": 2.3226,
      "step": 17690
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.717751530203827e-05,
      "loss": 2.2936,
      "step": 17700
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7096512463777702e-05,
      "loss": 2.2997,
      "step": 17710
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7015683210765664e-05,
      "loss": 2.2835,
      "step": 17720
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.693502771224428e-05,
      "loss": 2.3139,
      "step": 17730
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.6854546137091932e-05,
      "loss": 2.3089,
      "step": 17740
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.6774238653822814e-05,
      "loss": 2.3076,
      "step": 17750
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.6694105430586583e-05,
      "loss": 2.2998,
      "step": 17760
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.6614146635168025e-05,
      "loss": 2.3056,
      "step": 17770
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.6534362434986782e-05,
      "loss": 2.3181,
      "step": 17780
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.64547529970968e-05,
      "loss": 2.2939,
      "step": 17790
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.637531848818623e-05,
      "loss": 2.2714,
      "step": 17800
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.6296059074576863e-05,
      "loss": 2.2896,
      "step": 17810
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.6216974922223882e-05,
      "loss": 2.3024,
      "step": 17820
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.6138066196715573e-05,
      "loss": 2.2837,
      "step": 17830
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.6059333063272796e-05,
      "loss": 2.3029,
      "step": 17840
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.598077568674887e-05,
      "loss": 2.3029,
      "step": 17850
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.5902394231629026e-05,
      "loss": 2.2956,
      "step": 17860
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.5824188862030198e-05,
      "loss": 2.3181,
      "step": 17870
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.574615974170057e-05,
      "loss": 2.3091,
      "step": 17880
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.566830703401937e-05,
      "loss": 2.2924,
      "step": 17890
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.5590630901996385e-05,
      "loss": 2.2766,
      "step": 17900
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.5513131508271695e-05,
      "loss": 2.3049,
      "step": 17910
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.5435809015115354e-05,
      "loss": 2.2889,
      "step": 17920
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.535866358442698e-05,
      "loss": 2.3061,
      "step": 17930
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.528169537773545e-05,
      "loss": 2.3004,
      "step": 17940
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.520490455619863e-05,
      "loss": 2.2653,
      "step": 17950
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.512829128060288e-05,
      "loss": 2.306,
      "step": 17960
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.5051855711362905e-05,
      "loss": 2.2967,
      "step": 17970
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.4975598008521218e-05,
      "loss": 2.3043,
      "step": 17980
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.4899518331748019e-05,
      "loss": 2.2996,
      "step": 17990
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.482361684034067e-05,
      "loss": 2.3013,
      "step": 18000
    },
    {
      "epoch": 0.83,
      "eval_accuracy": 0.5487760253723954,
      "eval_loss": 2.1599340438842773,
      "eval_runtime": 10.3838,
      "eval_samples_per_second": 126.832,
      "eval_steps_per_second": 1.059,
      "step": 18000
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4747893693223513e-05,
      "loss": 2.2948,
      "step": 18010
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4672349048947409e-05,
      "loss": 2.3178,
      "step": 18020
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4596983065689463e-05,
      "loss": 2.3007,
      "step": 18030
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4521795901252777e-05,
      "loss": 2.3079,
      "step": 18040
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4446787713065935e-05,
      "loss": 2.3128,
      "step": 18050
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4371958658182871e-05,
      "loss": 2.3104,
      "step": 18060
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4297308893282369e-05,
      "loss": 2.3258,
      "step": 18070
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4222838574667819e-05,
      "loss": 2.3231,
      "step": 18080
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4148547858266958e-05,
      "loss": 2.3075,
      "step": 18090
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4074436899631383e-05,
      "loss": 2.3009,
      "step": 18100
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.400050585393633e-05,
      "loss": 2.3045,
      "step": 18110
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.3926754875980374e-05,
      "loss": 2.2964,
      "step": 18120
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.3853184120184993e-05,
      "loss": 2.296,
      "step": 18130
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.3779793740594371e-05,
      "loss": 2.2795,
      "step": 18140
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.3706583890874958e-05,
      "loss": 2.2873,
      "step": 18150
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.363355472431529e-05,
      "loss": 2.2935,
      "step": 18160
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.3560706393825474e-05,
      "loss": 2.2991,
      "step": 18170
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.3488039051937084e-05,
      "loss": 2.2907,
      "step": 18180
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.3415552850802648e-05,
      "loss": 2.3092,
      "step": 18190
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.3343247942195513e-05,
      "loss": 2.3221,
      "step": 18200
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.3271124477509333e-05,
      "loss": 2.2877,
      "step": 18210
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.3199182607757888e-05,
      "loss": 2.2859,
      "step": 18220
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.312742248357477e-05,
      "loss": 2.2861,
      "step": 18230
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.305584425521299e-05,
      "loss": 2.3274,
      "step": 18240
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.2984448072544686e-05,
      "loss": 2.2981,
      "step": 18250
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.291323408506082e-05,
      "loss": 2.3004,
      "step": 18260
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.2842202441870965e-05,
      "loss": 2.2897,
      "step": 18270
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.2771353291702759e-05,
      "loss": 2.3131,
      "step": 18280
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.2700686782901838e-05,
      "loss": 2.3092,
      "step": 18290
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.2630203063431356e-05,
      "loss": 2.2759,
      "step": 18300
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.2559902280871794e-05,
      "loss": 2.3059,
      "step": 18310
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.2489784582420527e-05,
      "loss": 2.3305,
      "step": 18320
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.2419850114891662e-05,
      "loss": 2.2879,
      "step": 18330
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.2350099024715567e-05,
      "loss": 2.2835,
      "step": 18340
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.2280531457938727e-05,
      "loss": 2.3077,
      "step": 18350
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.2211147560223313e-05,
      "loss": 2.2912,
      "step": 18360
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.214194747684697e-05,
      "loss": 2.2876,
      "step": 18370
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.2072931352702387e-05,
      "loss": 2.3263,
      "step": 18380
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.2004099332297191e-05,
      "loss": 2.3033,
      "step": 18390
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.1935451559753441e-05,
      "loss": 2.2875,
      "step": 18400
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.1866988178807426e-05,
      "loss": 2.3081,
      "step": 18410
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.1798709332809366e-05,
      "loss": 2.2874,
      "step": 18420
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.173061516472309e-05,
      "loss": 2.2866,
      "step": 18430
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1662705817125774e-05,
      "loss": 2.3109,
      "step": 18440
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1594981432207563e-05,
      "loss": 2.3099,
      "step": 18450
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.152744215177136e-05,
      "loss": 2.3134,
      "step": 18460
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.146008811723247e-05,
      "loss": 2.3128,
      "step": 18470
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1392919469618369e-05,
      "loss": 2.2952,
      "step": 18480
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1325936349568278e-05,
      "loss": 2.3005,
      "step": 18490
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1259138897333054e-05,
      "loss": 2.3217,
      "step": 18500
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1192527252774732e-05,
      "loss": 2.3002,
      "step": 18510
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1126101555366342e-05,
      "loss": 2.3062,
      "step": 18520
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1059861944191518e-05,
      "loss": 2.3026,
      "step": 18530
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.0993808557944341e-05,
      "loss": 2.2716,
      "step": 18540
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.0927941534928899e-05,
      "loss": 2.3138,
      "step": 18550
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.0862261013059082e-05,
      "loss": 2.2864,
      "step": 18560
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.079676712985832e-05,
      "loss": 2.2994,
      "step": 18570
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.0731460022459227e-05,
      "loss": 2.2817,
      "step": 18580
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.066633982760331e-05,
      "loss": 2.3024,
      "step": 18590
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.060140668164079e-05,
      "loss": 2.3003,
      "step": 18600
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.0536660720530156e-05,
      "loss": 2.2977,
      "step": 18610
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.0472102079838052e-05,
      "loss": 2.2878,
      "step": 18620
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.0407730894738821e-05,
      "loss": 2.3066,
      "step": 18630
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.0343547300014344e-05,
      "loss": 2.3014,
      "step": 18640
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.0279551430053747e-05,
      "loss": 2.2983,
      "step": 18650
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.0215743418853053e-05,
      "loss": 2.2866,
      "step": 18660
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.0152123400014979e-05,
      "loss": 2.2904,
      "step": 18670
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.008869150674856e-05,
      "loss": 2.3142,
      "step": 18680
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.002544787186902e-05,
      "loss": 2.3045,
      "step": 18690
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.962392627797301e-06,
      "loss": 2.297,
      "step": 18700
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.89952590655998e-06,
      "loss": 2.3033,
      "step": 18710
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.836847839788832e-06,
      "loss": 2.2794,
      "step": 18720
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.774358558720653e-06,
      "loss": 2.2866,
      "step": 18730
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.71205819419696e-06,
      "loss": 2.3037,
      "step": 18740
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.649946876663706e-06,
      "loss": 2.3129,
      "step": 18750
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.588024736170986e-06,
      "loss": 2.3076,
      "step": 18760
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.526291902372853e-06,
      "loss": 2.3311,
      "step": 18770
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.46474850452691e-06,
      "loss": 2.3084,
      "step": 18780
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.403394671494204e-06,
      "loss": 2.2871,
      "step": 18790
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.34223053173875e-06,
      "loss": 2.2848,
      "step": 18800
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.281256213327516e-06,
      "loss": 2.305,
      "step": 18810
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.220471843929901e-06,
      "loss": 2.3053,
      "step": 18820
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.15987755081763e-06,
      "loss": 2.3283,
      "step": 18830
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.099473460864461e-06,
      "loss": 2.3151,
      "step": 18840
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.039259700545866e-06,
      "loss": 2.3139,
      "step": 18850
    },
    {
      "epoch": 0.86,
      "learning_rate": 8.979236395938828e-06,
      "loss": 2.3077,
      "step": 18860
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.919403672721527e-06,
      "loss": 2.2889,
      "step": 18870
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.859761656173127e-06,
      "loss": 2.2996,
      "step": 18880
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.800310471173446e-06,
      "loss": 2.3225,
      "step": 18890
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.741050242202774e-06,
      "loss": 2.306,
      "step": 18900
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.681981093341529e-06,
      "loss": 2.2895,
      "step": 18910
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.623103148270096e-06,
      "loss": 2.3268,
      "step": 18920
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.56441653026846e-06,
      "loss": 2.3031,
      "step": 18930
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.505921362216052e-06,
      "loss": 2.3062,
      "step": 18940
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.447617766591387e-06,
      "loss": 2.299,
      "step": 18950
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.389505865471915e-06,
      "loss": 2.3101,
      "step": 18960
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.331585780533657e-06,
      "loss": 2.3182,
      "step": 18970
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.273857633051064e-06,
      "loss": 2.2893,
      "step": 18980
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.216321543896632e-06,
      "loss": 2.3137,
      "step": 18990
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.15897763354081e-06,
      "loss": 2.2957,
      "step": 19000
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.101826022051561e-06,
      "loss": 2.2961,
      "step": 19010
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.044866829094311e-06,
      "loss": 2.3183,
      "step": 19020
    },
    {
      "epoch": 0.87,
      "learning_rate": 7.988100173931501e-06,
      "loss": 2.2996,
      "step": 19030
    },
    {
      "epoch": 0.87,
      "learning_rate": 7.931526175422466e-06,
      "loss": 2.2866,
      "step": 19040
    },
    {
      "epoch": 0.87,
      "learning_rate": 7.875144952023195e-06,
      "loss": 2.3124,
      "step": 19050
    },
    {
      "epoch": 0.87,
      "learning_rate": 7.818956621785989e-06,
      "loss": 2.3046,
      "step": 19060
    },
    {
      "epoch": 0.87,
      "learning_rate": 7.762961302359273e-06,
      "loss": 2.2859,
      "step": 19070
    },
    {
      "epoch": 0.87,
      "learning_rate": 7.707159110987328e-06,
      "loss": 2.2784,
      "step": 19080
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.651550164510124e-06,
      "loss": 2.2708,
      "step": 19090
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.596134579362923e-06,
      "loss": 2.2869,
      "step": 19100
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.5409124715762e-06,
      "loss": 2.3068,
      "step": 19110
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.485883956775252e-06,
      "loss": 2.2944,
      "step": 19120
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.431049150180102e-06,
      "loss": 2.2998,
      "step": 19130
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.376408166605098e-06,
      "loss": 2.3286,
      "step": 19140
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.321961120458842e-06,
      "loss": 2.289,
      "step": 19150
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.2677081257437886e-06,
      "loss": 2.2907,
      "step": 19160
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.213649296056157e-06,
      "loss": 2.2979,
      "step": 19170
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.1597847445855515e-06,
      "loss": 2.2939,
      "step": 19180
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.106114584114831e-06,
      "loss": 2.3037,
      "step": 19190
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.052638927019817e-06,
      "loss": 2.291,
      "step": 19200
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.999357885269109e-06,
      "loss": 2.3114,
      "step": 19210
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.946271570423768e-06,
      "loss": 2.2876,
      "step": 19220
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.893380093637158e-06,
      "loss": 2.2974,
      "step": 19230
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.840683565654682e-06,
      "loss": 2.3158,
      "step": 19240
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.7881820968135336e-06,
      "loss": 2.2808,
      "step": 19250
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.735875797042535e-06,
      "loss": 2.3028,
      "step": 19260
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.683764775861823e-06,
      "loss": 2.2873,
      "step": 19270
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.631849142382674e-06,
      "loss": 2.309,
      "step": 19280
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.580129005307234e-06,
      "loss": 2.2927,
      "step": 19290
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.528604472928346e-06,
      "loss": 2.293,
      "step": 19300
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.477275653129233e-06,
      "loss": 2.3311,
      "step": 19310
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.426142653383427e-06,
      "loss": 2.3267,
      "step": 19320
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.375205580754351e-06,
      "loss": 2.2962,
      "step": 19330
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.324464541895247e-06,
      "loss": 2.3228,
      "step": 19340
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.273919643048865e-06,
      "loss": 2.3013,
      "step": 19350
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.223570990047323e-06,
      "loss": 2.2951,
      "step": 19360
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.173418688311772e-06,
      "loss": 2.276,
      "step": 19370
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.123462842852257e-06,
      "loss": 2.2913,
      "step": 19380
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.073703558267518e-06,
      "loss": 2.2927,
      "step": 19390
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.024140938744693e-06,
      "loss": 2.2891,
      "step": 19400
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.974775088059115e-06,
      "loss": 2.2829,
      "step": 19410
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.9256061095742e-06,
      "loss": 2.2801,
      "step": 19420
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.8766341062410635e-06,
      "loss": 2.3173,
      "step": 19430
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.827859180598427e-06,
      "loss": 2.2969,
      "step": 19440
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.779281434772376e-06,
      "loss": 2.2965,
      "step": 19450
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.730900970476105e-06,
      "loss": 2.3169,
      "step": 19460
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.682717889009781e-06,
      "loss": 2.3081,
      "step": 19470
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.634732291260225e-06,
      "loss": 2.3215,
      "step": 19480
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.5869442777008345e-06,
      "loss": 2.2931,
      "step": 19490
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.5393539483912234e-06,
      "loss": 2.2812,
      "step": 19500
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.491961402977164e-06,
      "loss": 2.2991,
      "step": 19510
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.444766740690244e-06,
      "loss": 2.2931,
      "step": 19520
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.3977700603477736e-06,
      "loss": 2.3059,
      "step": 19530
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.350971460352472e-06,
      "loss": 2.2928,
      "step": 19540
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.304371038692313e-06,
      "loss": 2.2998,
      "step": 19550
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.25796889294039e-06,
      "loss": 2.2951,
      "step": 19560
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.211765120254574e-06,
      "loss": 2.307,
      "step": 19570
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.16575981737738e-06,
      "loss": 2.3008,
      "step": 19580
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.1199530806358085e-06,
      "loss": 2.2797,
      "step": 19590
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.074345005941039e-06,
      "loss": 2.3225,
      "step": 19600
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.02893568878835e-06,
      "loss": 2.3083,
      "step": 19610
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.983725224256808e-06,
      "loss": 2.3095,
      "step": 19620
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.938713707009124e-06,
      "loss": 2.2676,
      "step": 19630
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.893901231291487e-06,
      "loss": 2.2795,
      "step": 19640
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.849287890933263e-06,
      "loss": 2.3209,
      "step": 19650
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.8048737793469415e-06,
      "loss": 2.3125,
      "step": 19660
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.760658989527778e-06,
      "loss": 2.311,
      "step": 19670
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.716643614053773e-06,
      "loss": 2.2761,
      "step": 19680
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.672827745085306e-06,
      "loss": 2.3034,
      "step": 19690
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.629211474365081e-06,
      "loss": 2.3094,
      "step": 19700
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.585794893217854e-06,
      "loss": 2.3131,
      "step": 19710
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.542578092550276e-06,
      "loss": 2.3131,
      "step": 19720
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.499561162850663e-06,
      "loss": 2.301,
      "step": 19730
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.4567441941889e-06,
      "loss": 2.2857,
      "step": 19740
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.414127276216107e-06,
      "loss": 2.2717,
      "step": 19750
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.371710498164627e-06,
      "loss": 2.3076,
      "step": 19760
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.329493948847663e-06,
      "loss": 2.306,
      "step": 19770
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.287477716659216e-06,
      "loss": 2.3287,
      "step": 19780
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.245661889573838e-06,
      "loss": 2.3002,
      "step": 19790
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.2040465551465014e-06,
      "loss": 2.3036,
      "step": 19800
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.16263180051234e-06,
      "loss": 2.3191,
      "step": 19810
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.121417712386544e-06,
      "loss": 2.3023,
      "step": 19820
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.080404377064129e-06,
      "loss": 2.3016,
      "step": 19830
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.039591880419746e-06,
      "loss": 2.3256,
      "step": 19840
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.998980307907574e-06,
      "loss": 2.2932,
      "step": 19850
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.958569744561047e-06,
      "loss": 2.3295,
      "step": 19860
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.918360274992772e-06,
      "loss": 2.3096,
      "step": 19870
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.878351983394246e-06,
      "loss": 2.2909,
      "step": 19880
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.838544953535772e-06,
      "loss": 2.3159,
      "step": 19890
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.7989392687662106e-06,
      "loss": 2.312,
      "step": 19900
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.7595350120128937e-06,
      "loss": 2.3168,
      "step": 19910
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.720332265781357e-06,
      "loss": 2.2988,
      "step": 19920
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.6813311121552175e-06,
      "loss": 2.3071,
      "step": 19930
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.6425316327959978e-06,
      "loss": 2.2792,
      "step": 19940
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.603933908942958e-06,
      "loss": 2.2842,
      "step": 19950
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.5655380214129063e-06,
      "loss": 2.2876,
      "step": 19960
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.5273440506000453e-06,
      "loss": 2.3033,
      "step": 19970
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.4893520764757824e-06,
      "loss": 2.3115,
      "step": 19980
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.451562178588619e-06,
      "loss": 2.3205,
      "step": 19990
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.4139744360639047e-06,
      "loss": 2.3075,
      "step": 20000
    },
    {
      "epoch": 0.92,
      "eval_accuracy": 0.5488688041410504,
      "eval_loss": 2.159656524658203,
      "eval_runtime": 10.5788,
      "eval_samples_per_second": 124.494,
      "eval_steps_per_second": 1.04,
      "step": 20000
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.376588927603741e-06,
      "loss": 2.2854,
      "step": 20010
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.3394057314867554e-06,
      "loss": 2.3114,
      "step": 20020
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.3024249255680152e-06,
      "loss": 2.3001,
      "step": 20030
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.2656465872787702e-06,
      "loss": 2.3023,
      "step": 20040
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.2290707936263766e-06,
      "loss": 2.3016,
      "step": 20050
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.192697621194063e-06,
      "loss": 2.2972,
      "step": 20060
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.1565271461408084e-06,
      "loss": 2.2982,
      "step": 20070
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.1205594442012433e-06,
      "loss": 2.2858,
      "step": 20080
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.084794590685336e-06,
      "loss": 2.3109,
      "step": 20090
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.049232660478385e-06,
      "loss": 2.2748,
      "step": 20100
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.0138737280407724e-06,
      "loss": 2.3026,
      "step": 20110
    },
    {
      "epoch": 0.92,
      "learning_rate": 2.9787178674078765e-06,
      "loss": 2.3137,
      "step": 20120
    },
    {
      "epoch": 0.92,
      "learning_rate": 2.943765152189837e-06,
      "loss": 2.2967,
      "step": 20130
    },
    {
      "epoch": 0.92,
      "learning_rate": 2.9090156555714675e-06,
      "loss": 2.2888,
      "step": 20140
    },
    {
      "epoch": 0.92,
      "learning_rate": 2.8744694503120564e-06,
      "loss": 2.3102,
      "step": 20150
    },
    {
      "epoch": 0.92,
      "learning_rate": 2.840126608745286e-06,
      "loss": 2.2931,
      "step": 20160
    },
    {
      "epoch": 0.92,
      "learning_rate": 2.8059872027789593e-06,
      "loss": 2.3183,
      "step": 20170
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.7720513038949735e-06,
      "loss": 2.2839,
      "step": 20180
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.738318983149113e-06,
      "loss": 2.3115,
      "step": 20190
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.704790311170868e-06,
      "loss": 2.3032,
      "step": 20200
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.671465358163383e-06,
      "loss": 2.3031,
      "step": 20210
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.638344193903197e-06,
      "loss": 2.3108,
      "step": 20220
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.60542688774017e-06,
      "loss": 2.3263,
      "step": 20230
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.572713508597324e-06,
      "loss": 2.345,
      "step": 20240
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.5402041249707133e-06,
      "loss": 2.2901,
      "step": 20250
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.5078988049292206e-06,
      "loss": 2.3061,
      "step": 20260
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.4757976161144836e-06,
      "loss": 2.286,
      "step": 20270
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.443900625740714e-06,
      "loss": 2.2841,
      "step": 20280
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.4122079005945763e-06,
      "loss": 2.3022,
      "step": 20290
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.380719507035023e-06,
      "loss": 2.2793,
      "step": 20300
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.349435510993192e-06,
      "loss": 2.3001,
      "step": 20310
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.3183559779722418e-06,
      "loss": 2.2981,
      "step": 20320
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.287480973047229e-06,
      "loss": 2.317,
      "step": 20330
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.2568105608649417e-06,
      "loss": 2.3118,
      "step": 20340
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.22634480564381e-06,
      "loss": 2.3064,
      "step": 20350
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.19608377117374e-06,
      "loss": 2.2939,
      "step": 20360
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.166027520815972e-06,
      "loss": 2.2939,
      "step": 20370
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.1361761175030195e-06,
      "loss": 2.3047,
      "step": 20380
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.106529623738407e-06,
      "loss": 2.3041,
      "step": 20390
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.077088101596658e-06,
      "loss": 2.309,
      "step": 20400
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.0478516127231397e-06,
      "loss": 2.3226,
      "step": 20410
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.0188202183338613e-06,
      "loss": 2.2901,
      "step": 20420
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.989993979215454e-06,
      "loss": 2.3097,
      "step": 20430
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.9613729557249605e-06,
      "loss": 2.281,
      "step": 20440
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.932957207789732e-06,
      "loss": 2.2964,
      "step": 20450
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.9047467949073416e-06,
      "loss": 2.2927,
      "step": 20460
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.8767417761453855e-06,
      "loss": 2.3154,
      "step": 20470
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.8489422101414467e-06,
      "loss": 2.2862,
      "step": 20480
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.8213481551028867e-06,
      "loss": 2.3085,
      "step": 20490
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.7939596688067772e-06,
      "loss": 2.2972,
      "step": 20500
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.7667768085997793e-06,
      "loss": 2.2874,
      "step": 20510
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.739799631397998e-06,
      "loss": 2.2767,
      "step": 20520
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.7130281936868608e-06,
      "loss": 2.2985,
      "step": 20530
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.6864625515210287e-06,
      "loss": 2.2929,
      "step": 20540
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.6601027605242624e-06,
      "loss": 2.287,
      "step": 20550
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.633948875889313e-06,
      "loss": 2.3101,
      "step": 20560
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.6080009523777862e-06,
      "loss": 2.2839,
      "step": 20570
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.5822590443200446e-06,
      "loss": 2.2993,
      "step": 20580
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.556723205615085e-06,
      "loss": 2.3081,
      "step": 20590
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.5313934897304595e-06,
      "loss": 2.2871,
      "step": 20600
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.5062699497020993e-06,
      "loss": 2.2979,
      "step": 20610
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.4813526381342702e-06,
      "loss": 2.2898,
      "step": 20620
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.456641607199416e-06,
      "loss": 2.301,
      "step": 20630
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.4321369086380488e-06,
      "loss": 2.2861,
      "step": 20640
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.4078385937587035e-06,
      "loss": 2.2985,
      "step": 20650
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.383746713437728e-06,
      "loss": 2.2992,
      "step": 20660
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.3598613181192821e-06,
      "loss": 2.2902,
      "step": 20670
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.3361824578151494e-06,
      "loss": 2.3184,
      "step": 20680
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.3127101821046927e-06,
      "loss": 2.3046,
      "step": 20690
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.2894445401346989e-06,
      "loss": 2.2876,
      "step": 20700
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.2663855806193114e-06,
      "loss": 2.3064,
      "step": 20710
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.2435333518398984e-06,
      "loss": 2.2963,
      "step": 20720
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.220887901645018e-06,
      "loss": 2.2857,
      "step": 20730
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.1984492774502088e-06,
      "loss": 2.3012,
      "step": 20740
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.1762175262380215e-06,
      "loss": 2.3118,
      "step": 20750
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.1541926945577653e-06,
      "loss": 2.3125,
      "step": 20760
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.1323748285255953e-06,
      "loss": 2.279,
      "step": 20770
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.1107639738242248e-06,
      "loss": 2.3304,
      "step": 20780
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.0893601757029913e-06,
      "loss": 2.3059,
      "step": 20790
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.0681634789776351e-06,
      "loss": 2.2903,
      "step": 20800
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.0471739280303316e-06,
      "loss": 2.2834,
      "step": 20810
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.0263915668094593e-06,
      "loss": 2.2779,
      "step": 20820
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.005816438829621e-06,
      "loss": 2.3038,
      "step": 20830
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.854485871715003e-07,
      "loss": 2.3111,
      "step": 20840
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.652880544817612e-07,
      "loss": 2.3159,
      "step": 20850
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.453348829730258e-07,
      "loss": 2.3179,
      "step": 20860
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.255891144236861e-07,
      "loss": 2.3159,
      "step": 20870
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.060507901778925e-07,
      "loss": 2.3088,
      "step": 20880
    },
    {
      "epoch": 0.96,
      "learning_rate": 8.867199511454316e-07,
      "loss": 2.3019,
      "step": 20890
    },
    {
      "epoch": 0.96,
      "learning_rate": 8.675966378016931e-07,
      "loss": 2.3064,
      "step": 20900
    },
    {
      "epoch": 0.96,
      "learning_rate": 8.486808901874921e-07,
      "loss": 2.3027,
      "step": 20910
    },
    {
      "epoch": 0.96,
      "learning_rate": 8.299727479090801e-07,
      "loss": 2.3103,
      "step": 20920
    },
    {
      "epoch": 0.96,
      "learning_rate": 8.114722501379679e-07,
      "loss": 2.3127,
      "step": 20930
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.931794356109579e-07,
      "loss": 2.2974,
      "step": 20940
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.750943426299562e-07,
      "loss": 2.2706,
      "step": 20950
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.572170090619612e-07,
      "loss": 2.2955,
      "step": 20960
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.395474723389306e-07,
      "loss": 2.2929,
      "step": 20970
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.220857694577698e-07,
      "loss": 2.2928,
      "step": 20980
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.048319369801881e-07,
      "loss": 2.2958,
      "step": 20990
    },
    {
      "epoch": 0.96,
      "learning_rate": 6.877860110326651e-07,
      "loss": 2.2842,
      "step": 21000
    },
    {
      "epoch": 0.96,
      "learning_rate": 6.709480273063507e-07,
      "loss": 2.2989,
      "step": 21010
    },
    {
      "epoch": 0.96,
      "learning_rate": 6.543180210570099e-07,
      "loss": 2.2944,
      "step": 21020
    },
    {
      "epoch": 0.96,
      "learning_rate": 6.37896027104945e-07,
      "loss": 2.2812,
      "step": 21030
    },
    {
      "epoch": 0.96,
      "learning_rate": 6.216820798348955e-07,
      "loss": 2.2956,
      "step": 21040
    },
    {
      "epoch": 0.97,
      "learning_rate": 6.056762131960048e-07,
      "loss": 2.2813,
      "step": 21050
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.898784607017205e-07,
      "loss": 2.2941,
      "step": 21060
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.74288855429761e-07,
      "loss": 2.3113,
      "step": 21070
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.589074300219932e-07,
      "loss": 2.2998,
      "step": 21080
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.437342166844106e-07,
      "loss": 2.2744,
      "step": 21090
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.287692471870331e-07,
      "loss": 2.3032,
      "step": 21100
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.140125528638628e-07,
      "loss": 2.294,
      "step": 21110
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.994641646128062e-07,
      "loss": 2.2723,
      "step": 21120
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.851241128956186e-07,
      "loss": 2.3043,
      "step": 21130
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.7099242773783746e-07,
      "loss": 2.2919,
      "step": 21140
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.5706913872871626e-07,
      "loss": 2.2847,
      "step": 21150
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.4335427502114613e-07,
      "loss": 2.3028,
      "step": 21160
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.298478653316451e-07,
      "loss": 2.295,
      "step": 21170
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.16549937940236e-07,
      "loss": 2.3075,
      "step": 21180
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.0346052069042404e-07,
      "loss": 2.286,
      "step": 21190
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.9057964098915265e-07,
      "loss": 2.3007,
      "step": 21200
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.779073258066923e-07,
      "loss": 2.308,
      "step": 21210
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.6544360167664047e-07,
      "loss": 2.3119,
      "step": 21220
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.5318849469582194e-07,
      "loss": 2.3065,
      "step": 21230
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.411420305242774e-07,
      "loss": 2.3026,
      "step": 21240
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.293042343851638e-07,
      "loss": 2.314,
      "step": 21250
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.176751310647319e-07,
      "loss": 2.3058,
      "step": 21260
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.0625474491227104e-07,
      "loss": 2.3121,
      "step": 21270
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.950430998400644e-07,
      "loss": 2.3067,
      "step": 21280
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.840402193232894e-07,
      "loss": 2.3006,
      "step": 21290
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.7324612640005077e-07,
      "loss": 2.2893,
      "step": 21300
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.626608436712696e-07,
      "loss": 2.2836,
      "step": 21310
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.5228439330065025e-07,
      "loss": 2.2999,
      "step": 21320
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.421167970146576e-07,
      "loss": 2.3,
      "step": 21330
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.321580761024289e-07,
      "loss": 2.305,
      "step": 21340
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.2240825141577327e-07,
      "loss": 2.3297,
      "step": 21350
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.128673433690942e-07,
      "loss": 2.3106,
      "step": 21360
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.0353537193936735e-07,
      "loss": 2.2859,
      "step": 21370
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.9441235666609602e-07,
      "loss": 2.2827,
      "step": 21380
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.8549831665124474e-07,
      "loss": 2.3143,
      "step": 21390
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.767932705592723e-07,
      "loss": 2.2779,
      "step": 21400
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.682972366169655e-07,
      "loss": 2.2856,
      "step": 21410
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.6001023261354997e-07,
      "loss": 2.3009,
      "step": 21420
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.5193227590053484e-07,
      "loss": 2.2864,
      "step": 21430
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.4406338339173487e-07,
      "loss": 2.3091,
      "step": 21440
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.3640357156321504e-07,
      "loss": 2.2979,
      "step": 21450
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.2895285645326828e-07,
      "loss": 2.3053,
      "step": 21460
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.217112536623821e-07,
      "loss": 2.3144,
      "step": 21470
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.146787783531833e-07,
      "loss": 2.29,
      "step": 21480
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.0785544525044878e-07,
      "loss": 2.2866,
      "step": 21490
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.0124126864100581e-07,
      "loss": 2.2878,
      "step": 21500
    },
    {
      "epoch": 0.99,
      "learning_rate": 9.483626237379862e-08,
      "loss": 2.3039,
      "step": 21510
    },
    {
      "epoch": 0.99,
      "learning_rate": 8.864043985975511e-08,
      "loss": 2.296,
      "step": 21520
    },
    {
      "epoch": 0.99,
      "learning_rate": 8.265381407185357e-08,
      "loss": 2.3034,
      "step": 21530
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.687639754501152e-08,
      "loss": 2.2954,
      "step": 21540
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.130820237611913e-08,
      "loss": 2.2945,
      "step": 21550
    },
    {
      "epoch": 0.99,
      "learning_rate": 6.594924022400583e-08,
      "loss": 2.2996,
      "step": 21560
    },
    {
      "epoch": 0.99,
      "learning_rate": 6.07995223093849e-08,
      "loss": 2.2881,
      "step": 21570
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.5859059414853364e-08,
      "loss": 2.3122,
      "step": 21580
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.112786188486984e-08,
      "loss": 2.3026,
      "step": 21590
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.660593962572124e-08,
      "loss": 2.3117,
      "step": 21600
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.229330210552273e-08,
      "loss": 2.2909,
      "step": 21610
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.818995835417338e-08,
      "loss": 2.3051,
      "step": 21620
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.42959169633339e-08,
      "loss": 2.3006,
      "step": 21630
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.0611186086459963e-08,
      "loss": 2.2787,
      "step": 21640
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.7135773438702328e-08,
      "loss": 2.2938,
      "step": 21650
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.3869686296984495e-08,
      "loss": 2.3208,
      "step": 21660
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.0812931499913922e-08,
      "loss": 2.306,
      "step": 21670
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.796551544777092e-08,
      "loss": 2.2907,
      "step": 21680
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.532744410257525e-08,
      "loss": 2.3234,
      "step": 21690
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.2898722987952916e-08,
      "loss": 2.2921,
      "step": 21700
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.067935718922497e-08,
      "loss": 2.2851,
      "step": 21710
    },
    {
      "epoch": 1.0,
      "learning_rate": 8.669351353363108e-09,
      "loss": 2.319,
      "step": 21720
    },
    {
      "epoch": 1.0,
      "learning_rate": 6.868709688945263e-09,
      "loss": 2.3035,
      "step": 21730
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.27743596622221e-09,
      "loss": 2.3272,
      "step": 21740
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.8955335170065554e-09,
      "loss": 2.3093,
      "step": 21750
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.723005234794851e-09,
      "loss": 2.3023,
      "step": 21760
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.7598535746232713e-09,
      "loss": 2.3055,
      "step": 21770
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.006080553167532e-09,
      "loss": 2.2972,
      "step": 21780
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.616877487095828e-10,
      "loss": 2.3003,
      "step": 21790
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.266763011043004e-10,
      "loss": 2.3028,
      "step": 21800
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.0469118016942503e-12,
      "loss": 2.2949,
      "step": 21810
    },
    {
      "epoch": 1.0,
      "step": 21811,
      "total_flos": 1.3200227455536307e+20,
      "train_loss": 2.360168211507445,
      "train_runtime": 23329.1279,
      "train_samples_per_second": 119.668,
      "train_steps_per_second": 0.935
    }
  ],
  "logging_steps": 10,
  "max_steps": 21811,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 10000,
  "total_flos": 1.3200227455536307e+20,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}