{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.7408779403593257,
  "global_step": 4000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 0.0009999978838190456,
      "loss": 2.9794,
      "step": 10
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0009999915352940948,
      "loss": 2.3885,
      "step": 20
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.000999980954478887,
      "loss": 2.3057,
      "step": 30
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.000999966141462985,
      "loss": 2.2692,
      "step": 40
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.000999947096371777,
      "loss": 2.2576,
      "step": 50
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0009999238193664748,
      "loss": 2.2388,
      "step": 60
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0009998963106441117,
      "loss": 2.2523,
      "step": 70
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0009998645704375414,
      "loss": 2.218,
      "step": 80
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.000999828599015436,
      "loss": 2.2457,
      "step": 90
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0009997883966822835,
      "loss": 2.198,
      "step": 100
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0009997439637783859,
      "loss": 2.2013,
      "step": 110
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.000999695300679855,
      "loss": 2.1765,
      "step": 120
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0009996424077986109,
      "loss": 2.1741,
      "step": 130
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.000999585285582377,
      "loss": 2.1898,
      "step": 140
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0009995239345146772,
      "loss": 2.1466,
      "step": 150
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0009994583551148314,
      "loss": 2.1423,
      "step": 160
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0009993885479379506,
      "loss": 2.1451,
      "step": 170
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.000999314513574934,
      "loss": 2.202,
      "step": 180
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0009992362526524616,
      "loss": 2.1208,
      "step": 190
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0009991537658329906,
      "loss": 2.1591,
      "step": 200
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.000999067053814749,
      "loss": 2.1788,
      "step": 210
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0009989761173317304,
      "loss": 2.147,
      "step": 220
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.000998880957153687,
      "loss": 2.1249,
      "step": 230
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.000998781574086123,
      "loss": 2.165,
      "step": 240
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.000998677968970289,
      "loss": 2.1428,
      "step": 250
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0009985701426831735,
      "loss": 2.1384,
      "step": 260
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0009984580961374964,
      "loss": 2.1585,
      "step": 270
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0009983418302817008,
      "loss": 2.1156,
      "step": 280
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0009982213460999448,
      "loss": 2.0811,
      "step": 290
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.000998096644612094,
      "loss": 2.1081,
      "step": 300
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0009979677268737118,
      "loss": 2.1246,
      "step": 310
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0009978345939760515,
      "loss": 2.1229,
      "step": 320
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.000997697247046046,
      "loss": 2.1033,
      "step": 330
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0009975556872462994,
      "loss": 2.0931,
      "step": 340
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.000997409915775076,
      "loss": 2.1206,
      "step": 350
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0009972599338662915,
      "loss": 2.0537,
      "step": 360
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0009971057427895012,
      "loss": 2.0762,
      "step": 370
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0009969473438498897,
      "loss": 2.0883,
      "step": 380
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0009967847383882604,
      "loss": 2.1174,
      "step": 390
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0009966179277810239,
      "loss": 2.1111,
      "step": 400
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0009964469134401855,
      "loss": 2.1288,
      "step": 410
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0009962716968133346,
      "loss": 2.0967,
      "step": 420
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0009960922793836318,
      "loss": 2.1216,
      "step": 430
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0009959086626697955,
      "loss": 2.0924,
      "step": 440
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0009957208482260908,
      "loss": 2.0809,
      "step": 450
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0009955288376423152,
      "loss": 2.1082,
      "step": 460
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0009953326325437852,
      "loss": 2.0885,
      "step": 470
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0009951322345913224,
      "loss": 2.1133,
      "step": 480
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0009949276454812408,
      "loss": 2.0844,
      "step": 490
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00099471886694533,
      "loss": 2.0796,
      "step": 500
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0009945059007508434,
      "loss": 2.1255,
      "step": 510
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0009942887487004804,
      "loss": 2.0913,
      "step": 520
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0009940674126323733,
      "loss": 2.1003,
      "step": 530
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0009938418944200709,
      "loss": 2.0541,
      "step": 540
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0009936121959725223,
      "loss": 2.0523,
      "step": 550
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0009933783192340618,
      "loss": 2.1225,
      "step": 560
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0009931402661843911,
      "loss": 2.0446,
      "step": 570
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.000992898038838564,
      "loss": 2.0921,
      "step": 580
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0009926516392469674,
      "loss": 2.1081,
      "step": 590
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0009924010694953064,
      "loss": 2.0734,
      "step": 600
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0009921463317045843,
      "loss": 2.0652,
      "step": 610
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0009918874280310862,
      "loss": 2.0818,
      "step": 620
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0009916243606663605,
      "loss": 2.0776,
      "step": 630
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0009913571318371994,
      "loss": 2.1025,
      "step": 640
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0009910857438056215,
      "loss": 2.066,
      "step": 650
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0009908101988688512,
      "loss": 2.0575,
      "step": 660
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0009905304993593008,
      "loss": 2.1269,
      "step": 670
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0009902466476445486,
      "loss": 2.0518,
      "step": 680
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0009899586461273218,
      "loss": 2.0698,
      "step": 690
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.000989666497245473,
      "loss": 2.0988,
      "step": 700
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0009893702034719624,
      "loss": 2.0986,
      "step": 710
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0009890697673148345,
      "loss": 2.0237,
      "step": 720
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0009887651913171986,
      "loss": 2.0027,
      "step": 730
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0009884564780572064,
      "loss": 2.0563,
      "step": 740
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0009881436301480305,
      "loss": 2.0624,
      "step": 750
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.000987826650237842,
      "loss": 2.0926,
      "step": 760
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.000987505541009788,
      "loss": 2.0585,
      "step": 770
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0009871803051819696,
      "loss": 2.0494,
      "step": 780
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0009868509455074183,
      "loss": 2.0106,
      "step": 790
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0009865174647740729,
      "loss": 2.0861,
      "step": 800
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0009861798658047556,
      "loss": 2.0478,
      "step": 810
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0009858381514571484,
      "loss": 2.0469,
      "step": 820
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.000985492324623769,
      "loss": 2.0671,
      "step": 830
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0009851423882319458,
      "loss": 2.0808,
      "step": 840
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0009847883452437937,
      "loss": 2.0331,
      "step": 850
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0009844301986561893,
      "loss": 2.0295,
      "step": 860
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.000984067951500744,
      "loss": 2.0873,
      "step": 870
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00098370160684378,
      "loss": 2.1038,
      "step": 880
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0009833311677863042,
      "loss": 2.0337,
      "step": 890
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0009829566374639801,
      "loss": 2.0407,
      "step": 900
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0009825780190471042,
      "loss": 2.1049,
      "step": 910
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.000982195315740576,
      "loss": 2.0475,
      "step": 920
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0009818085307838741,
      "loss": 2.0624,
      "step": 930
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.000981417667451026,
      "loss": 2.0714,
      "step": 940
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0009810227290505816,
      "loss": 2.0947,
      "step": 950
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0009806237189255859,
      "loss": 2.0591,
      "step": 960
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0009802206404535489,
      "loss": 2.0301,
      "step": 970
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.000979813497046419,
      "loss": 2.0556,
      "step": 980
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0009794022921505523,
      "loss": 2.0753,
      "step": 990
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.000978987029246685,
      "loss": 2.0898,
      "step": 1000
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0009785677118499029,
      "loss": 2.0464,
      "step": 1010
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0009781443435096116,
      "loss": 2.0828,
      "step": 1020
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0009777169278095074,
      "loss": 2.1137,
      "step": 1030
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0009772854683675462,
      "loss": 2.0167,
      "step": 1040
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.000976849968835913,
      "loss": 2.07,
      "step": 1050
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0009764104329009909,
      "loss": 2.0409,
      "step": 1060
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0009759668642833304,
      "loss": 2.015,
      "step": 1070
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0009755192667376173,
      "loss": 2.0175,
      "step": 1080
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0009750676440526411,
      "loss": 2.0773,
      "step": 1090
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0009746120000512632,
      "loss": 2.0245,
      "step": 1100
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0009741523385903841,
      "loss": 2.094,
      "step": 1110
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0009736886635609112,
      "loss": 2.0506,
      "step": 1120
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0009732209788877258,
      "loss": 2.0287,
      "step": 1130
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0009727492885296489,
      "loss": 2.1162,
      "step": 1140
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0009722735964794099,
      "loss": 2.1096,
      "step": 1150
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0009717939067636099,
      "loss": 2.0621,
      "step": 1160
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0009713102234426903,
      "loss": 2.0796,
      "step": 1170
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0009708225506108965,
      "loss": 2.0565,
      "step": 1180
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0009703308923962447,
      "loss": 2.0669,
      "step": 1190
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0009698352529604857,
      "loss": 2.0638,
      "step": 1200
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0009693356364990705,
      "loss": 2.0358,
      "step": 1210
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0009688320472411143,
      "loss": 2.0859,
      "step": 1220
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0009683244894493613,
      "loss": 2.0932,
      "step": 1230
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0009678129674201479,
      "loss": 2.0129,
      "step": 1240
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0009672974854833669,
      "loss": 2.055,
      "step": 1250
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0009667780480024304,
      "loss": 2.0665,
      "step": 1260
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0009662546593742334,
      "loss": 2.0488,
      "step": 1270
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0009657273240291159,
      "loss": 2.0543,
      "step": 1280
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0009651960464308261,
      "loss": 2.0418,
      "step": 1290
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0009646608310764819,
      "loss": 2.033,
      "step": 1300
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0009641216824965338,
      "loss": 2.1034,
      "step": 1310
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0009635786052547253,
      "loss": 2.0866,
      "step": 1320
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0009630316039480556,
      "loss": 2.0607,
      "step": 1330
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0009624806832067394,
      "loss": 2.0457,
      "step": 1340
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0009619258476941686,
      "loss": 2.032,
      "step": 1350
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.000961367102106873,
      "loss": 2.0519,
      "step": 1360
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0009608044511744791,
      "loss": 2.0449,
      "step": 1370
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0009602378996596721,
      "loss": 1.9949,
      "step": 1380
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0009596674523581539,
      "loss": 2.0394,
      "step": 1390
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0009590931140986035,
      "loss": 2.0386,
      "step": 1400
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0009585148897426354,
      "loss": 2.0254,
      "step": 1410
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0009579327841847593,
      "loss": 2.0238,
      "step": 1420
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.000957346802352338,
      "loss": 2.0509,
      "step": 1430
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0009567569492055456,
      "loss": 2.0004,
      "step": 1440
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0009561632297373263,
      "loss": 2.0203,
      "step": 1450
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0009555656489733513,
      "loss": 2.0182,
      "step": 1460
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.000954964211971977,
      "loss": 1.9754,
      "step": 1470
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0009543589238242012,
      "loss": 2.0374,
      "step": 1480
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.000953749789653621,
      "loss": 2.0367,
      "step": 1490
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.000953136814616389,
      "loss": 2.0866,
      "step": 1500
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0009525200039011694,
      "loss": 2.0083,
      "step": 1510
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0009518993627290948,
      "loss": 2.0525,
      "step": 1520
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0009512748963537212,
      "loss": 2.0636,
      "step": 1530
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.000950646610060984,
      "loss": 2.0522,
      "step": 1540
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0009500145091691532,
      "loss": 2.05,
      "step": 1550
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0009493785990287882,
      "loss": 1.9887,
      "step": 1560
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0009487388850226926,
      "loss": 2.0309,
      "step": 1570
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.000948095372565869,
      "loss": 1.9954,
      "step": 1580
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0009474480671054726,
      "loss": 2.0078,
      "step": 1590
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0009467969741207652,
      "loss": 2.0395,
      "step": 1600
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0009461420991230693,
      "loss": 2.0415,
      "step": 1610
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0009454834476557207,
      "loss": 2.0308,
      "step": 1620
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0009448210252940223,
      "loss": 2.0826,
      "step": 1630
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0009441548376451963,
      "loss": 2.0424,
      "step": 1640
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0009434848903483373,
      "loss": 2.0125,
      "step": 1650
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0009428111890743639,
      "loss": 2.0139,
      "step": 1660
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0009421337395259717,
      "loss": 2.0682,
      "step": 1670
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0009414525474375837,
      "loss": 2.0577,
      "step": 1680
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0009407676185753029,
      "loss": 2.0262,
      "step": 1690
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0009400789587368632,
      "loss": 2.0515,
      "step": 1700
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0009393865737515794,
      "loss": 2.0398,
      "step": 1710
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0009386904694802997,
      "loss": 2.0146,
      "step": 1720
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0009379906518153543,
      "loss": 2.0438,
      "step": 1730
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0009372871266805063,
      "loss": 2.0377,
      "step": 1740
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.000936579900030902,
      "loss": 2.0789,
      "step": 1750
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0009358689778530193,
      "loss": 2.0201,
      "step": 1760
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0009351543661646185,
      "loss": 2.0114,
      "step": 1770
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0009344360710146898,
      "loss": 2.0242,
      "step": 1780
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0009337140984834034,
      "loss": 2.0436,
      "step": 1790
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0009329884546820572,
      "loss": 2.0452,
      "step": 1800
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.000932259145753026,
      "loss": 2.0254,
      "step": 1810
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0009315261778697083,
      "loss": 2.0409,
      "step": 1820
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0009307895572364746,
      "loss": 2.0301,
      "step": 1830
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0009300492900886154,
      "loss": 2.0078,
      "step": 1840
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0009293053826922873,
      "loss": 1.9851,
      "step": 1850
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0009285578413444613,
      "loss": 1.9947,
      "step": 1860
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0009278066723728682,
      "loss": 2.0331,
      "step": 1870
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0009270518821359461,
      "loss": 2.0058,
      "step": 1880
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0009262934770227858,
      "loss": 2.05,
      "step": 1890
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0009255314634530771,
      "loss": 2.0444,
      "step": 1900
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0009247658478770543,
      "loss": 2.0045,
      "step": 1910
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.000923996636775442,
      "loss": 2.0211,
      "step": 1920
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0009232238366593997,
      "loss": 2.0124,
      "step": 1930
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0009224474540704671,
      "loss": 2.0067,
      "step": 1940
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0009216674955805079,
      "loss": 2.0247,
      "step": 1950
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0009208839677916557,
      "loss": 2.0314,
      "step": 1960
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0009200968773362568,
      "loss": 2.067,
      "step": 1970
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0009193062308768145,
      "loss": 2.0168,
      "step": 1980
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0009185120351059326,
      "loss": 2.0649,
      "step": 1990
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0009177142967462591,
      "loss": 2.0208,
      "step": 2000
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0009169130225504289,
      "loss": 2.016,
      "step": 2010
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0009161082193010066,
      "loss": 2.0331,
      "step": 2020
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0009152998938104296,
      "loss": 2.0757,
      "step": 2030
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0009144880529209498,
      "loss": 2.0093,
      "step": 2040
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0009136727035045765,
      "loss": 1.954,
      "step": 2050
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0009128538524630167,
      "loss": 2.0062,
      "step": 2060
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0009120315067276187,
      "loss": 2.0179,
      "step": 2070
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0009112056732593119,
      "loss": 2.0301,
      "step": 2080
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0009103763590485488,
      "loss": 2.0223,
      "step": 2090
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0009095435711152449,
      "loss": 2.0208,
      "step": 2100
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00090870731650872,
      "loss": 2.0492,
      "step": 2110
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0009078676023076385,
      "loss": 1.9917,
      "step": 2120
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0009070244356199492,
      "loss": 2.0235,
      "step": 2130
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0009061778235828253,
      "loss": 2.01,
      "step": 2140
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0009054129328980229,
      "loss": 1.9999,
      "step": 2150
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.000904559794464287,
      "loss": 2.0022,
      "step": 2160
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0009037032315436229,
      "loss": 2.0004,
      "step": 2170
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0009028432513865994,
      "loss": 1.9936,
      "step": 2180
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.000901979861272711,
      "loss": 2.0098,
      "step": 2190
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0009011130685103166,
      "loss": 1.9829,
      "step": 2200
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0009002428804365775,
      "loss": 2.0135,
      "step": 2210
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0008993693044173954,
      "loss": 2.0526,
      "step": 2220
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0008984923478473499,
      "loss": 2.0145,
      "step": 2230
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0008976120181496365,
      "loss": 2.0364,
      "step": 2240
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0008967283227760028,
      "loss": 2.019,
      "step": 2250
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.000895841269206686,
      "loss": 1.9874,
      "step": 2260
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0008949508649503495,
      "loss": 2.0699,
      "step": 2270
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0008940571175440196,
      "loss": 2.0407,
      "step": 2280
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0008931600345530213,
      "loss": 2.0176,
      "step": 2290
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0008922596235709142,
      "loss": 2.0235,
      "step": 2300
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0008913558922194287,
      "loss": 2.0268,
      "step": 2310
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.000890448848148401,
      "loss": 2.0279,
      "step": 2320
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0008895384990357087,
      "loss": 2.0073,
      "step": 2330
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0008886248525872056,
      "loss": 2.0262,
      "step": 2340
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0008877079165366566,
      "loss": 2.0112,
      "step": 2350
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.000886787698645672,
      "loss": 2.042,
      "step": 2360
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0008858642067036424,
      "loss": 2.0129,
      "step": 2370
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0008849374485276715,
      "loss": 2.0264,
      "step": 2380
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.000884007431962512,
      "loss": 2.0211,
      "step": 2390
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0008830741648804967,
      "loss": 2.0175,
      "step": 2400
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0008821376551814739,
      "loss": 2.0423,
      "step": 2410
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0008811979107927396,
      "loss": 2.0112,
      "step": 2420
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0008802549396689705,
      "loss": 2.0422,
      "step": 2430
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0008793087497921566,
      "loss": 2.0075,
      "step": 2440
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0008784544334882426,
      "loss": 1.9958,
      "step": 2450
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.000877502150068553,
      "loss": 2.0348,
      "step": 2460
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0008765466711974225,
      "loss": 2.0232,
      "step": 2470
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0008755880049627157,
      "loss": 1.9953,
      "step": 2480
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0008746261594792776,
      "loss": 2.0223,
      "step": 2490
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0008736611428888644,
      "loss": 2.0332,
      "step": 2500
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0008726929633600753,
      "loss": 2.0075,
      "step": 2510
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0008717216290882824,
      "loss": 2.002,
      "step": 2520
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0008707471482955622,
      "loss": 2.0594,
      "step": 2530
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0008697695292306253,
      "loss": 2.0214,
      "step": 2540
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0008687887801687473,
      "loss": 1.994,
      "step": 2550
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0008678049094116979,
      "loss": 2.0338,
      "step": 2560
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0008668179252876714,
      "loss": 2.0176,
      "step": 2570
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0008658278361512161,
      "loss": 2.0013,
      "step": 2580
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0008648346503831627,
      "loss": 2.0141,
      "step": 2590
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0008638383763905546,
      "loss": 1.9655,
      "step": 2600
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0008628390226065761,
      "loss": 2.039,
      "step": 2610
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0008618365974904808,
      "loss": 1.9418,
      "step": 2620
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0008608311095275205,
      "loss": 2.011,
      "step": 2630
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0008598225672288731,
      "loss": 2.0174,
      "step": 2640
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0008588109791315707,
      "loss": 1.984,
      "step": 2650
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.000857796353798427,
      "loss": 2.0437,
      "step": 2660
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0008567786998179654,
      "loss": 2.0255,
      "step": 2670
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0008557580258043459,
      "loss": 2.0067,
      "step": 2680
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0008547343403972918,
      "loss": 2.0201,
      "step": 2690
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0008537076522620175,
      "loss": 2.0258,
      "step": 2700
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0008526779700891545,
      "loss": 1.9977,
      "step": 2710
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0008516453025946781,
      "loss": 1.9835,
      "step": 2720
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0008506096585198333,
      "loss": 2.0428,
      "step": 2730
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0008495710466310613,
      "loss": 1.9554,
      "step": 2740
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0008485294757199248,
      "loss": 2.0497,
      "step": 2750
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0008474849546030339,
      "loss": 2.001,
      "step": 2760
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0008464374921219714,
      "loss": 1.9824,
      "step": 2770
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0008453870971432179,
      "loss": 2.0378,
      "step": 2780
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0008443337785580769,
      "loss": 1.9756,
      "step": 2790
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0008432775452825992,
      "loss": 2.0563,
      "step": 2800
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0008422184062575077,
      "loss": 2.0115,
      "step": 2810
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0008411563704481217,
      "loss": 2.0301,
      "step": 2820
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0008400914468442814,
      "loss": 1.9912,
      "step": 2830
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0008390236444602706,
      "loss": 2.0669,
      "step": 2840
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0008379529723347417,
      "loss": 2.0138,
      "step": 2850
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0008368794395306386,
      "loss": 2.0486,
      "step": 2860
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0008358030551351199,
      "loss": 2.0342,
      "step": 2870
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0008347238282594823,
      "loss": 2.0474,
      "step": 2880
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0008336417680390829,
      "loss": 2.006,
      "step": 2890
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0008325568836332633,
      "loss": 1.99,
      "step": 2900
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00083146918422527,
      "loss": 2.0148,
      "step": 2910
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0008303786790221778,
      "loss": 2.0381,
      "step": 2920
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0008292853772548126,
      "loss": 2.0591,
      "step": 2930
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0008281892881776714,
      "loss": 2.0238,
      "step": 2940
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0008270904210688459,
      "loss": 1.9939,
      "step": 2950
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0008259887852299427,
      "loss": 1.9941,
      "step": 2960
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0008248843899860045,
      "loss": 2.0193,
      "step": 2970
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0008237772446854325,
      "loss": 2.0166,
      "step": 2980
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0008226673586999058,
      "loss": 2.0219,
      "step": 2990
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0008215547414243025,
      "loss": 2.0078,
      "step": 3000
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0008204394022766208,
      "loss": 2.0022,
      "step": 3010
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0008193213506978983,
      "loss": 2.0009,
      "step": 3020
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0008182005961521331,
      "loss": 2.0197,
      "step": 3030
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0008170771481262027,
      "loss": 2.0101,
      "step": 3040
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0008159510161297846,
      "loss": 2.0307,
      "step": 3050
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0008148222096952748,
      "loss": 2.0075,
      "step": 3060
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0008136907383777081,
      "loss": 2.0415,
      "step": 3070
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.000812556611754677,
      "loss": 2.0031,
      "step": 3080
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0008114198394262497,
      "loss": 2.0025,
      "step": 3090
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0008102804310148905,
      "loss": 1.983,
      "step": 3100
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0008091383961653765,
      "loss": 2.0056,
      "step": 3110
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0008079937445447175,
      "loss": 1.9679,
      "step": 3120
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0008068464858420732,
      "loss": 1.9897,
      "step": 3130
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0008056966297686717,
      "loss": 1.9833,
      "step": 3140
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0008045441860577273,
      "loss": 2.0088,
      "step": 3150
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0008033891644643576,
      "loss": 1.9985,
      "step": 3160
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0008022315747655011,
      "loss": 2.013,
      "step": 3170
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0008010714267598354,
      "loss": 1.9902,
      "step": 3180
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0007999087302676926,
      "loss": 2.0133,
      "step": 3190
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0007987434951309774,
      "loss": 2.0092,
      "step": 3200
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0007975757312130836,
      "loss": 1.9894,
      "step": 3210
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0007964054483988103,
      "loss": 1.9742,
      "step": 3220
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.000795232656594278,
      "loss": 2.016,
      "step": 3230
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0007940573657268458,
      "loss": 2.0386,
      "step": 3240
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0007928795857450259,
      "loss": 1.9602,
      "step": 3250
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0007916993266184007,
      "loss": 1.9681,
      "step": 3260
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0007905165983375379,
      "loss": 2.0326,
      "step": 3270
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0007893314109139058,
      "loss": 1.9958,
      "step": 3280
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0007881437743797882,
      "loss": 2.0077,
      "step": 3290
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0007869536987882008,
      "loss": 1.9813,
      "step": 3300
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0007857611942128044,
      "loss": 1.9538,
      "step": 3310
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0007845662707478214,
      "loss": 1.9919,
      "step": 3320
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0007833689385079485,
      "loss": 1.9922,
      "step": 3330
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0007821692076282725,
      "loss": 1.9973,
      "step": 3340
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0007809670882641842,
      "loss": 2.0231,
      "step": 3350
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0007797625905912915,
      "loss": 1.9971,
      "step": 3360
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0007785557248053351,
      "loss": 1.9934,
      "step": 3370
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0007773465011221002,
      "loss": 1.9855,
      "step": 3380
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0007761349297773315,
      "loss": 2.0421,
      "step": 3390
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0007749210210266457,
      "loss": 1.9795,
      "step": 3400
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0007737047851454451,
      "loss": 2.0006,
      "step": 3410
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0007724862324288309,
      "loss": 2.011,
      "step": 3420
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0007712653731915147,
      "loss": 1.9969,
      "step": 3430
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0007700422177677331,
      "loss": 1.9581,
      "step": 3440
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0007688167765111592,
      "loss": 1.9858,
      "step": 3450
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0007675890597948141,
      "loss": 1.9945,
      "step": 3460
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0007663590780109814,
      "loss": 2.0116,
      "step": 3470
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0007651268415711168,
      "loss": 1.9831,
      "step": 3480
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0007638923609057616,
      "loss": 2.0113,
      "step": 3490
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0007626556464644538,
      "loss": 1.9927,
      "step": 3500
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0007614167087156393,
      "loss": 2.0209,
      "step": 3510
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0007601755581465844,
      "loss": 1.9941,
      "step": 3520
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0007589322052632855,
      "loss": 2.0443,
      "step": 3530
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0007576866605903812,
      "loss": 1.975,
      "step": 3540
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0007564389346710638,
      "loss": 1.9655,
      "step": 3550
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0007551890380669879,
      "loss": 2.0034,
      "step": 3560
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0007539369813581836,
      "loss": 1.953,
      "step": 3570
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.000752682775142965,
      "loss": 2.0162,
      "step": 3580
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0007514264300378411,
      "loss": 2.0096,
      "step": 3590
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0007501679566774267,
      "loss": 2.0431,
      "step": 3600
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0007489073657143506,
      "loss": 1.961,
      "step": 3610
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0007476446678191676,
      "loss": 2.0179,
      "step": 3620
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0007463798736802665,
      "loss": 1.9885,
      "step": 3630
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0007451129940037804,
      "loss": 1.9955,
      "step": 3640
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.000743844039513496,
      "loss": 1.9671,
      "step": 3650
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0007425730209507622,
      "loss": 1.9866,
      "step": 3660
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0007412999490744008,
      "loss": 1.9508,
      "step": 3670
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0007400248346606129,
      "loss": 2.0261,
      "step": 3680
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0007387476885028902,
      "loss": 1.9946,
      "step": 3690
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0007374685214119221,
      "loss": 1.969,
      "step": 3700
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0007361873442155049,
      "loss": 1.9802,
      "step": 3710
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0007349041677584496,
      "loss": 1.9739,
      "step": 3720
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0007336190029024907,
      "loss": 1.964,
      "step": 3730
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0007323318605261933,
      "loss": 2.0009,
      "step": 3740
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0007310427515248626,
      "loss": 1.9825,
      "step": 3750
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0007297516868104501,
      "loss": 1.9744,
      "step": 3760
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.000728458677311462,
      "loss": 2.0197,
      "step": 3770
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0007271637339728667,
      "loss": 2.0016,
      "step": 3780
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0007258668677560019,
      "loss": 1.9721,
      "step": 3790
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0007245680896384818,
      "loss": 1.9591,
      "step": 3800
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0007232674106141048,
      "loss": 1.9762,
      "step": 3810
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0007219648416927592,
      "loss": 2.0026,
      "step": 3820
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0007206603939003314,
      "loss": 1.9876,
      "step": 3830
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0007193540782786117,
      "loss": 1.9991,
      "step": 3840
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0007180459058852011,
      "loss": 1.9897,
      "step": 3850
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0007167358877934176,
      "loss": 2.0581,
      "step": 3860
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0007154240350922025,
      "loss": 1.9588,
      "step": 3870
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0007141103588860265,
      "loss": 2.0151,
      "step": 3880
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.000712794870294796,
      "loss": 1.9494,
      "step": 3890
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0007114775804537586,
      "loss": 1.9936,
      "step": 3900
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0007101585005134088,
      "loss": 2.0016,
      "step": 3910
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0007088376416393944,
      "loss": 1.9628,
      "step": 3920
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0007075150150124208,
      "loss": 1.9955,
      "step": 3930
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0007061906318281568,
      "loss": 1.9731,
      "step": 3940
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0007048645032971407,
      "loss": 1.9554,
      "step": 3950
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0007035366406446839,
      "loss": 1.985,
      "step": 3960
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0007022070551107774,
      "loss": 1.9961,
      "step": 3970
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0007008757579499954,
      "loss": 2.0112,
      "step": 3980
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0006995427604314004,
      "loss": 2.0086,
      "step": 3990
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0006982080738384487,
      "loss": 2.019,
      "step": 4000
    }
  ],
  "max_steps": 10798,
  "num_train_epochs": 2,
  "total_flos": 3.794844442012156e+18,
  "trial_name": null,
  "trial_params": null
}