{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.23500658018424517,
  "eval_steps": 500,
  "global_step": 10000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 0.0003999999863730145,
      "loss": 8.118,
      "step": 10
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0003999999454920598,
      "loss": 6.3286,
      "step": 20
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0003999998773571415,
      "loss": 6.1146,
      "step": 30
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00039999978196826884,
      "loss": 5.8777,
      "step": 40
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0003999996593254548,
      "loss": 5.5376,
      "step": 50
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00039999950942871623,
      "loss": 5.2922,
      "step": 60
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0003999993322780734,
      "loss": 5.0843,
      "step": 70
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00039999912787355056,
      "loss": 4.8923,
      "step": 80
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0003999988962151755,
      "loss": 4.7488,
      "step": 90
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0003999986373029798,
      "loss": 4.6656,
      "step": 100
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00039999835113699875,
      "loss": 4.5166,
      "step": 110
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00039999803771727135,
      "loss": 4.4329,
      "step": 120
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0003999976970438403,
      "loss": 4.408,
      "step": 130
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00039999732911675205,
      "loss": 4.2873,
      "step": 140
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00039999693393605676,
      "loss": 4.2245,
      "step": 150
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0003999965115018082,
      "loss": 4.2141,
      "step": 160
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0003999960618140639,
      "loss": 4.0998,
      "step": 170
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00039999558487288525,
      "loss": 4.0249,
      "step": 180
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00039999508067833727,
      "loss": 3.9842,
      "step": 190
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00039999454923048854,
      "loss": 3.9469,
      "step": 200
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0003999939905294116,
      "loss": 3.8331,
      "step": 210
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039999340457518245,
      "loss": 3.8387,
      "step": 220
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039999279136788106,
      "loss": 3.7919,
      "step": 230
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039999215090759095,
      "loss": 3.7312,
      "step": 240
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039999148319439935,
      "loss": 3.674,
      "step": 250
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039999078822839734,
      "loss": 3.6375,
      "step": 260
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999900660096795,
      "loss": 3.6078,
      "step": 270
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999893165383444,
      "loss": 3.5456,
      "step": 280
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039998853981449404,
      "loss": 3.4769,
      "step": 290
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039998773583823434,
      "loss": 3.4389,
      "step": 300
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039998690460967484,
      "loss": 3.389,
      "step": 310
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039998604612892876,
      "loss": 3.4085,
      "step": 320
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039998516039611317,
      "loss": 3.3085,
      "step": 330
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039998424741134877,
      "loss": 3.2635,
      "step": 340
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999833071747598,
      "loss": 3.2114,
      "step": 350
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999823396864746,
      "loss": 3.1907,
      "step": 360
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999813449466249,
      "loss": 3.1599,
      "step": 370
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999803229553463,
      "loss": 3.1176,
      "step": 380
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039997927371277804,
      "loss": 3.0683,
      "step": 390
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999781972190631,
      "loss": 3.0275,
      "step": 400
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999770934743482,
      "loss": 3.0021,
      "step": 410
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999759624787837,
      "loss": 2.9149,
      "step": 420
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999748042325237,
      "loss": 2.884,
      "step": 430
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999736187357261,
      "loss": 2.9275,
      "step": 440
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039997240598855247,
      "loss": 2.9095,
      "step": 450
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039997116599116796,
      "loss": 2.9409,
      "step": 460
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039996989874374167,
      "loss": 2.8867,
      "step": 470
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039996860424644623,
      "loss": 2.849,
      "step": 480
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039996728249945806,
      "loss": 2.8417,
      "step": 490
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039996593350295724,
      "loss": 2.8284,
      "step": 500
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039996455725712754,
      "loss": 2.7558,
      "step": 510
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039996315376215663,
      "loss": 2.7321,
      "step": 520
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999617230182358,
      "loss": 2.6914,
      "step": 530
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039996026502555975,
      "loss": 2.6755,
      "step": 540
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039995877978432745,
      "loss": 2.6632,
      "step": 550
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039995726729474113,
      "loss": 2.6723,
      "step": 560
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039995572755700695,
      "loss": 2.5948,
      "step": 570
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039995416057133477,
      "loss": 2.6085,
      "step": 580
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039995256633793805,
      "loss": 2.5718,
      "step": 590
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.000399950944857034,
      "loss": 2.5738,
      "step": 600
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039994929612884375,
      "loss": 2.5005,
      "step": 610
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999476201535918,
      "loss": 2.5343,
      "step": 620
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999459169315066,
      "loss": 2.547,
      "step": 630
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003999441864628203,
      "loss": 2.5294,
      "step": 640
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003999424287477686,
      "loss": 2.518,
      "step": 650
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039994064378659114,
      "loss": 2.5036,
      "step": 660
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039993883157953106,
      "loss": 2.4465,
      "step": 670
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003999369921268353,
      "loss": 2.4591,
      "step": 680
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039993512542875465,
      "loss": 2.4806,
      "step": 690
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003999332314855434,
      "loss": 2.4915,
      "step": 700
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003999313102974596,
      "loss": 2.4737,
      "step": 710
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039992936186476513,
      "loss": 2.4002,
      "step": 720
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003999273861877254,
      "loss": 2.4821,
      "step": 730
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003999253832666098,
      "loss": 2.445,
      "step": 740
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003999233531016911,
      "loss": 2.3891,
      "step": 750
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.000399921295693246,
      "loss": 2.3283,
      "step": 760
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039991921104155494,
      "loss": 2.4011,
      "step": 770
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003999170991469019,
      "loss": 2.4398,
      "step": 780
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039991496000957473,
      "loss": 2.4085,
      "step": 790
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039991279362986483,
      "loss": 2.3319,
      "step": 800
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039991060000806755,
      "loss": 2.3053,
      "step": 810
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039990837914448174,
      "loss": 2.3027,
      "step": 820
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039990613103941007,
      "loss": 2.4183,
      "step": 830
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003999038556931588,
      "loss": 2.3537,
      "step": 840
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003999015531060381,
      "loss": 2.3063,
      "step": 850
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003998992232783617,
      "loss": 2.3585,
      "step": 860
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039989686621044713,
      "loss": 2.3435,
      "step": 870
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039989448190261545,
      "loss": 2.3105,
      "step": 880
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003998920703551917,
      "loss": 2.3027,
      "step": 890
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003998896315685045,
      "loss": 2.2045,
      "step": 900
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003998871655428861,
      "loss": 2.2763,
      "step": 910
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003998846722786726,
      "loss": 2.2635,
      "step": 920
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039988215177620375,
      "loss": 2.2985,
      "step": 930
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039987960403582305,
      "loss": 2.2659,
      "step": 940
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039987702905787756,
      "loss": 2.2959,
      "step": 950
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039987442684271833,
      "loss": 2.2554,
      "step": 960
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003998717973906999,
      "loss": 2.2962,
      "step": 970
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039986914070218055,
      "loss": 2.2713,
      "step": 980
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039986645677752235,
      "loss": 2.2178,
      "step": 990
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.000399863745617091,
      "loss": 2.2294,
      "step": 1000
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.000399861007221256,
      "loss": 2.2203,
      "step": 1010
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039985824159039044,
      "loss": 2.2326,
      "step": 1020
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039985544872487124,
      "loss": 2.2135,
      "step": 1030
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039985262862507904,
      "loss": 2.1806,
      "step": 1040
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039984978129139803,
      "loss": 2.2396,
      "step": 1050
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039984690672421625,
      "loss": 2.1817,
      "step": 1060
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003998440049239254,
      "loss": 2.1996,
      "step": 1070
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039984107589092094,
      "loss": 2.139,
      "step": 1080
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003998381196256021,
      "loss": 2.1516,
      "step": 1090
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039983513612837146,
      "loss": 2.2708,
      "step": 1100
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003998321253996359,
      "loss": 2.2253,
      "step": 1110
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039982908743980543,
      "loss": 2.1869,
      "step": 1120
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039982602224929425,
      "loss": 2.1836,
      "step": 1130
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003998229298285199,
      "loss": 2.165,
      "step": 1140
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003998198101779038,
      "loss": 2.1514,
      "step": 1150
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039981666329787116,
      "loss": 2.2043,
      "step": 1160
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003998134891888507,
      "loss": 2.2431,
      "step": 1170
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.000399810287851275,
      "loss": 2.1645,
      "step": 1180
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003998070592855803,
      "loss": 2.0972,
      "step": 1190
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003998038034922066,
      "loss": 2.1676,
      "step": 1200
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039980052047159753,
      "loss": 2.1515,
      "step": 1210
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039979721022420043,
      "loss": 2.2242,
      "step": 1220
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039979387275046636,
      "loss": 2.1498,
      "step": 1230
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039979050805085027,
      "loss": 2.1727,
      "step": 1240
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039978711612581055,
      "loss": 2.1848,
      "step": 1250
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003997836969758094,
      "loss": 2.2028,
      "step": 1260
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039978025060131284,
      "loss": 2.1458,
      "step": 1270
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003997767770027904,
      "loss": 2.1835,
      "step": 1280
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003997732761807155,
      "loss": 2.1333,
      "step": 1290
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003997697481355652,
      "loss": 2.1516,
      "step": 1300
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003997661928678203,
      "loss": 2.1426,
      "step": 1310
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039976261037796514,
      "loss": 2.2161,
      "step": 1320
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.000399759000666488,
      "loss": 2.2368,
      "step": 1330
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039975536373388075,
      "loss": 2.1565,
      "step": 1340
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.000399751699580639,
      "loss": 2.083,
      "step": 1350
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039974800820726214,
      "loss": 2.0986,
      "step": 1360
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039974428961425304,
      "loss": 2.0592,
      "step": 1370
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039974054380211856,
      "loss": 2.0629,
      "step": 1380
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.000399736770771369,
      "loss": 2.0514,
      "step": 1390
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039973297052251877,
      "loss": 2.0839,
      "step": 1400
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039972914305608546,
      "loss": 2.1392,
      "step": 1410
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039972528837259075,
      "loss": 2.1073,
      "step": 1420
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003997214064725599,
      "loss": 2.0508,
      "step": 1430
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003997174973565219,
      "loss": 2.0333,
      "step": 1440
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003997135610250095,
      "loss": 2.0698,
      "step": 1450
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.000399709597478559,
      "loss": 2.0841,
      "step": 1460
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003997056067177106,
      "loss": 2.0923,
      "step": 1470
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003997015887430081,
      "loss": 2.064,
      "step": 1480
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039969754355499894,
      "loss": 2.0102,
      "step": 1490
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039969347115423445,
      "loss": 2.063,
      "step": 1500
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039968937154126965,
      "loss": 2.0544,
      "step": 1510
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.000399685244716663,
      "loss": 2.0805,
      "step": 1520
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039968109068097703,
      "loss": 2.0353,
      "step": 1530
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003996769094347777,
      "loss": 2.0498,
      "step": 1540
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003996727009786349,
      "loss": 2.0424,
      "step": 1550
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039966846531312196,
      "loss": 2.0201,
      "step": 1560
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003996642024388162,
      "loss": 2.006,
      "step": 1570
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003996599123562985,
      "loss": 2.0948,
      "step": 1580
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003996555950661534,
      "loss": 2.0324,
      "step": 1590
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003996512505689693,
      "loss": 2.046,
      "step": 1600
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003996468788653382,
      "loss": 2.0836,
      "step": 1610
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039964247995585573,
      "loss": 2.0208,
      "step": 1620
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003996380538411215,
      "loss": 2.061,
      "step": 1630
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003996336005217386,
      "loss": 2.0576,
      "step": 1640
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003996291199983138,
      "loss": 1.9996,
      "step": 1650
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039962461227145777,
      "loss": 1.9895,
      "step": 1660
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003996200773417847,
      "loss": 2.1121,
      "step": 1670
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039961551520991263,
      "loss": 2.1313,
      "step": 1680
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003996109258764632,
      "loss": 2.0451,
      "step": 1690
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003996063093420618,
      "loss": 2.0172,
      "step": 1700
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039960166560733745,
      "loss": 2.1171,
      "step": 1710
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.000399596994672923,
      "loss": 2.1142,
      "step": 1720
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003995922965394551,
      "loss": 2.0765,
      "step": 1730
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003995875712075738,
      "loss": 1.9818,
      "step": 1740
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003995828186779231,
      "loss": 2.0097,
      "step": 1750
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039957803895115056,
      "loss": 2.0391,
      "step": 1760
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039957323202790754,
      "loss": 2.0719,
      "step": 1770
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003995683979088491,
      "loss": 1.9692,
      "step": 1780
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.000399563536594634,
      "loss": 1.9804,
      "step": 1790
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003995586480859246,
      "loss": 2.0772,
      "step": 1800
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039955373238338715,
      "loss": 2.0121,
      "step": 1810
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003995487894876915,
      "loss": 2.1017,
      "step": 1820
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003995438193995111,
      "loss": 2.0021,
      "step": 1830
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003995388221195234,
      "loss": 2.0173,
      "step": 1840
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003995337976484093,
      "loss": 2.12,
      "step": 1850
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039952874598685347,
      "loss": 2.0556,
      "step": 1860
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003995236671355444,
      "loss": 2.1336,
      "step": 1870
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003995185610951739,
      "loss": 2.0388,
      "step": 1880
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003995134278664382,
      "loss": 2.0134,
      "step": 1890
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003995082674500364,
      "loss": 1.997,
      "step": 1900
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039950307984667197,
      "loss": 2.0323,
      "step": 1910
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039949786505705167,
      "loss": 2.0107,
      "step": 1920
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039949262308188615,
      "loss": 2.0288,
      "step": 1930
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003994873539218898,
      "loss": 1.9891,
      "step": 1940
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003994820575777807,
      "loss": 2.0481,
      "step": 1950
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003994767340502804,
      "loss": 2.0216,
      "step": 1960
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039947138334011446,
      "loss": 2.0167,
      "step": 1970
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039946600544801196,
      "loss": 1.9372,
      "step": 1980
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039946060037470587,
      "loss": 1.935,
      "step": 1990
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039945516812093254,
      "loss": 2.0051,
      "step": 2000
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003994497086874324,
      "loss": 2.087,
      "step": 2010
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003994442220749493,
      "loss": 1.9893,
      "step": 2020
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003994387082842309,
      "loss": 1.9887,
      "step": 2030
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003994331673160287,
      "loss": 1.9826,
      "step": 2040
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039942759917109755,
      "loss": 1.9206,
      "step": 2050
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003994220038501964,
      "loss": 2.0448,
      "step": 2060
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003994163813540876,
      "loss": 1.9772,
      "step": 2070
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003994107316835375,
      "loss": 2.0458,
      "step": 2080
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039940505483931573,
      "loss": 1.9609,
      "step": 2090
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039939935082219604,
      "loss": 1.9851,
      "step": 2100
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039939361963295573,
      "loss": 1.954,
      "step": 2110
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039938786127237576,
      "loss": 1.9591,
      "step": 2120
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003993820757412407,
      "loss": 1.9289,
      "step": 2130
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039937626304033914,
      "loss": 1.9337,
      "step": 2140
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.000399370423170463,
      "loss": 1.9348,
      "step": 2150
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003993645561324082,
      "loss": 2.0066,
      "step": 2160
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003993586619269742,
      "loss": 1.9672,
      "step": 2170
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039935274055496426,
      "loss": 2.0074,
      "step": 2180
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039934679201718515,
      "loss": 2.0142,
      "step": 2190
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039934081631444755,
      "loss": 1.9753,
      "step": 2200
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039933481344756587,
      "loss": 1.9794,
      "step": 2210
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003993287834173579,
      "loss": 2.01,
      "step": 2220
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003993227262246455,
      "loss": 1.9906,
      "step": 2230
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003993166418702541,
      "loss": 1.9185,
      "step": 2240
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003993105303550127,
      "loss": 1.919,
      "step": 2250
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003993043916797543,
      "loss": 1.9834,
      "step": 2260
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003992982258453152,
      "loss": 1.9505,
      "step": 2270
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003992920328525358,
      "loss": 1.9617,
      "step": 2280
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039928581270225986,
      "loss": 1.9801,
      "step": 2290
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003992795653953351,
      "loss": 1.9085,
      "step": 2300
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003992732909326128,
      "loss": 2.01,
      "step": 2310
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.000399266989314948,
      "loss": 2.0069,
      "step": 2320
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003992606605431994,
      "loss": 1.9633,
      "step": 2330
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003992543046182295,
      "loss": 2.0229,
      "step": 2340
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039924792154090433,
      "loss": 1.9732,
      "step": 2350
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039924151131209373,
      "loss": 1.8956,
      "step": 2360
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003992350739326712,
      "loss": 1.9605,
      "step": 2370
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003992286094035141,
      "loss": 1.9142,
      "step": 2380
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003992221177255031,
      "loss": 1.9233,
      "step": 2390
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039921559889952304,
      "loss": 1.95,
      "step": 2400
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003992090529264622,
      "loss": 1.9638,
      "step": 2410
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003992024798072125,
      "loss": 1.9341,
      "step": 2420
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003991958795426697,
      "loss": 1.9212,
      "step": 2430
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039918925213373326,
      "loss": 1.9003,
      "step": 2440
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003991825975813062,
      "loss": 1.91,
      "step": 2450
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003991759158862955,
      "loss": 1.9405,
      "step": 2460
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039916920704961156,
      "loss": 1.939,
      "step": 2470
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003991624710721686,
      "loss": 1.8843,
      "step": 2480
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039915570795488454,
      "loss": 1.9161,
      "step": 2490
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.000399148917698681,
      "loss": 1.912,
      "step": 2500
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039914210030448326,
      "loss": 2.029,
      "step": 2510
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039913525577322037,
      "loss": 1.9293,
      "step": 2520
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039912838410582495,
      "loss": 1.9556,
      "step": 2530
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003991214853032335,
      "loss": 1.9542,
      "step": 2540
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003991145593663861,
      "loss": 1.931,
      "step": 2550
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039910760629622644,
      "loss": 1.8945,
      "step": 2560
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003991006260937022,
      "loss": 1.9022,
      "step": 2570
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039909361875976445,
      "loss": 1.9475,
      "step": 2580
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039908658429536805,
      "loss": 1.9387,
      "step": 2590
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003990795227014716,
      "loss": 1.966,
      "step": 2600
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003990724339790375,
      "loss": 1.9118,
      "step": 2610
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003990653181290316,
      "loss": 1.9823,
      "step": 2620
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039905817515242364,
      "loss": 1.9006,
      "step": 2630
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.000399051005050187,
      "loss": 1.9616,
      "step": 2640
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039904380782329867,
      "loss": 2.0014,
      "step": 2650
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003990365834727395,
      "loss": 1.9463,
      "step": 2660
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039902933199949383,
      "loss": 1.9225,
      "step": 2670
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039902205340454995,
      "loss": 1.9477,
      "step": 2680
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039901474768889967,
      "loss": 1.8736,
      "step": 2690
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039900741485353856,
      "loss": 1.9268,
      "step": 2700
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039900005489946583,
      "loss": 1.9873,
      "step": 2710
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003989926678276844,
      "loss": 1.9226,
      "step": 2720
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.000398985253639201,
      "loss": 1.8946,
      "step": 2730
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039897781233502586,
      "loss": 1.9803,
      "step": 2740
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.000398970343916173,
      "loss": 1.9063,
      "step": 2750
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039896284838366025,
      "loss": 1.9195,
      "step": 2760
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003989553257385089,
      "loss": 1.8676,
      "step": 2770
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003989477759817442,
      "loss": 1.9615,
      "step": 2780
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003989401991143948,
      "loss": 1.9745,
      "step": 2790
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003989325951374933,
      "loss": 1.9975,
      "step": 2800
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039892496405207584,
      "loss": 1.906,
      "step": 2810
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003989173058591823,
      "loss": 1.8932,
      "step": 2820
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039890962055985635,
      "loss": 1.9158,
      "step": 2830
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039890190815514517,
      "loss": 1.9122,
      "step": 2840
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039889416864609975,
      "loss": 1.8768,
      "step": 2850
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003988864020337748,
      "loss": 1.9045,
      "step": 2860
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003988786083192286,
      "loss": 1.9719,
      "step": 2870
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003988707875035233,
      "loss": 1.8816,
      "step": 2880
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039886293958772456,
      "loss": 1.962,
      "step": 2890
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003988550645729018,
      "loss": 1.967,
      "step": 2900
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039884716246012817,
      "loss": 1.9085,
      "step": 2910
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003988392332504805,
      "loss": 1.8951,
      "step": 2920
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003988312769450394,
      "loss": 1.9603,
      "step": 2930
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003988232935448889,
      "loss": 1.9027,
      "step": 2940
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039881528305111695,
      "loss": 1.8541,
      "step": 2950
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003988072454648152,
      "loss": 1.9238,
      "step": 2960
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003987991807870789,
      "loss": 1.899,
      "step": 2970
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039879108901900697,
      "loss": 1.8818,
      "step": 2980
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003987829701617022,
      "loss": 1.8995,
      "step": 2990
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039877482421627077,
      "loss": 1.9103,
      "step": 3000
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003987666511838229,
      "loss": 1.8458,
      "step": 3010
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003987584510654722,
      "loss": 1.9115,
      "step": 3020
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039875022386233624,
      "loss": 1.9156,
      "step": 3030
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039874196957553594,
      "loss": 1.8478,
      "step": 3040
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039873368820619626,
      "loss": 1.8816,
      "step": 3050
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003987253797554457,
      "loss": 1.9044,
      "step": 3060
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003987170442244164,
      "loss": 1.9477,
      "step": 3070
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039870868161424416,
      "loss": 1.8782,
      "step": 3080
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003987002919260687,
      "loss": 1.9034,
      "step": 3090
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003986918751610333,
      "loss": 1.851,
      "step": 3100
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039868343132028476,
      "loss": 1.8951,
      "step": 3110
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003986749604049738,
      "loss": 1.8947,
      "step": 3120
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003986664624162548,
      "loss": 1.8742,
      "step": 3130
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003986579373552857,
      "loss": 1.8696,
      "step": 3140
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039864938522322826,
      "loss": 1.8318,
      "step": 3150
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039864080602124776,
      "loss": 1.8975,
      "step": 3160
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003986321997505135,
      "loss": 1.8946,
      "step": 3170
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003986235664121981,
      "loss": 1.9007,
      "step": 3180
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.000398614906007478,
      "loss": 1.8543,
      "step": 3190
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003986062185375335,
      "loss": 1.8906,
      "step": 3200
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039859750400354836,
      "loss": 1.8717,
      "step": 3210
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039858876240671004,
      "loss": 1.9091,
      "step": 3220
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039857999374820984,
      "loss": 1.8534,
      "step": 3230
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039857119802924265,
      "loss": 1.9424,
      "step": 3240
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039856237525100706,
      "loss": 1.8563,
      "step": 3250
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039855352541470535,
      "loss": 1.8508,
      "step": 3260
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003985446485215434,
      "loss": 1.7738,
      "step": 3270
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039853574457273104,
      "loss": 1.8507,
      "step": 3280
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003985268135694815,
      "loss": 1.87,
      "step": 3290
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039851785551301176,
      "loss": 1.8337,
      "step": 3300
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003985088704045426,
      "loss": 1.9019,
      "step": 3310
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039849985824529847,
      "loss": 1.892,
      "step": 3320
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039849081903650733,
      "loss": 1.8693,
      "step": 3330
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039848175277940105,
      "loss": 1.9654,
      "step": 3340
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003984726594752151,
      "loss": 1.9894,
      "step": 3350
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003984635391251885,
      "loss": 1.8574,
      "step": 3360
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003984543917305642,
      "loss": 1.8305,
      "step": 3370
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039844521729258864,
      "loss": 1.8939,
      "step": 3380
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039843601581251206,
      "loss": 1.8208,
      "step": 3390
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003984267872915883,
      "loss": 1.8424,
      "step": 3400
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039841753173107505,
      "loss": 1.8248,
      "step": 3410
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003984082491322334,
      "loss": 1.863,
      "step": 3420
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003983989394963284,
      "loss": 1.8603,
      "step": 3430
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039838960282462867,
      "loss": 1.9456,
      "step": 3440
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039838023911840644,
      "loss": 1.8169,
      "step": 3450
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003983708483789378,
      "loss": 1.8177,
      "step": 3460
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039836143060750225,
      "loss": 1.8531,
      "step": 3470
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039835198580538336,
      "loss": 1.9448,
      "step": 3480
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039834251397386807,
      "loss": 1.8522,
      "step": 3490
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039833301511424716,
      "loss": 1.8559,
      "step": 3500
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003983234892278149,
      "loss": 1.864,
      "step": 3510
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003983139363158696,
      "loss": 1.8782,
      "step": 3520
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039830435637971284,
      "loss": 1.8399,
      "step": 3530
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003982947494206501,
      "loss": 1.8912,
      "step": 3540
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039828511543999064,
      "loss": 1.8414,
      "step": 3550
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003982754544390471,
      "loss": 1.8746,
      "step": 3560
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039826576641913624,
      "loss": 1.8617,
      "step": 3570
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.000398256051381578,
      "loss": 1.8677,
      "step": 3580
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039824630932769643,
      "loss": 1.8652,
      "step": 3590
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003982365402588189,
      "loss": 1.8284,
      "step": 3600
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003982267441762767,
      "loss": 1.8363,
      "step": 3610
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003982169210814049,
      "loss": 1.8223,
      "step": 3620
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039820707097554183,
      "loss": 1.8814,
      "step": 3630
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039819719386002995,
      "loss": 1.889,
      "step": 3640
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003981872897362152,
      "loss": 1.8921,
      "step": 3650
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039817735860544714,
      "loss": 1.8233,
      "step": 3660
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003981674004690791,
      "loss": 1.8574,
      "step": 3670
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039815741532846807,
      "loss": 1.8394,
      "step": 3680
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003981474031849748,
      "loss": 1.8477,
      "step": 3690
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039813736403996355,
      "loss": 1.8046,
      "step": 3700
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003981272978948024,
      "loss": 1.8568,
      "step": 3710
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.000398117204750863,
      "loss": 1.8511,
      "step": 3720
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003981070846095208,
      "loss": 1.8222,
      "step": 3730
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039809693747215494,
      "loss": 1.8691,
      "step": 3740
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039808676334014806,
      "loss": 1.8915,
      "step": 3750
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003980765622148866,
      "loss": 1.8411,
      "step": 3760
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003980663340977607,
      "loss": 1.8555,
      "step": 3770
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039805607899016406,
      "loss": 1.8563,
      "step": 3780
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039804579689349436,
      "loss": 1.8344,
      "step": 3790
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003980354878091524,
      "loss": 1.8375,
      "step": 3800
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003980251517385433,
      "loss": 1.8031,
      "step": 3810
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003980147886830755,
      "loss": 1.8695,
      "step": 3820
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039800439864416106,
      "loss": 1.8657,
      "step": 3830
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039799398162321584,
      "loss": 1.8441,
      "step": 3840
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039798353762165946,
      "loss": 1.9097,
      "step": 3850
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039797306664091504,
      "loss": 1.9122,
      "step": 3860
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003979625686824095,
      "loss": 1.8143,
      "step": 3870
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003979520437475734,
      "loss": 1.859,
      "step": 3880
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003979414918378409,
      "loss": 1.8044,
      "step": 3890
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039793091295465004,
      "loss": 1.8097,
      "step": 3900
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003979203070994423,
      "loss": 1.8495,
      "step": 3910
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.000397909674273663,
      "loss": 1.8872,
      "step": 3920
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039789901447876097,
      "loss": 1.8748,
      "step": 3930
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039788832771618887,
      "loss": 1.7635,
      "step": 3940
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003978776139874031,
      "loss": 1.8179,
      "step": 3950
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003978668732938635,
      "loss": 1.8709,
      "step": 3960
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003978561056370337,
      "loss": 1.8009,
      "step": 3970
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039784531101838105,
      "loss": 1.8061,
      "step": 3980
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039783448943937646,
      "loss": 1.8609,
      "step": 3990
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003978236409014947,
      "loss": 1.8439,
      "step": 4000
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039781276540621393,
      "loss": 1.8459,
      "step": 4010
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003978018629550164,
      "loss": 1.8014,
      "step": 4020
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003977909335493876,
      "loss": 1.7949,
      "step": 4030
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003977799771908169,
      "loss": 1.8656,
      "step": 4040
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003977689938807973,
      "loss": 1.842,
      "step": 4050
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039775798362082557,
      "loss": 1.8599,
      "step": 4060
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039774694641240204,
      "loss": 1.8555,
      "step": 4070
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003977358822570308,
      "loss": 1.8812,
      "step": 4080
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039772479115621944,
      "loss": 1.8152,
      "step": 4090
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039771367311147943,
      "loss": 1.8578,
      "step": 4100
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039770252812432584,
      "loss": 1.7968,
      "step": 4110
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003976913561962774,
      "loss": 1.812,
      "step": 4120
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039768015732885646,
      "loss": 1.799,
      "step": 4130
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039766893152358906,
      "loss": 1.8426,
      "step": 4140
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039765767878200493,
      "loss": 1.8531,
      "step": 4150
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003976463991056376,
      "loss": 1.8041,
      "step": 4160
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003976350924960241,
      "loss": 1.8543,
      "step": 4170
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039762375895470507,
      "loss": 1.8657,
      "step": 4180
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003976123984832251,
      "loss": 1.8165,
      "step": 4190
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039760101108313216,
      "loss": 1.8312,
      "step": 4200
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003975895967559781,
      "loss": 1.8181,
      "step": 4210
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003975781555033183,
      "loss": 1.8071,
      "step": 4220
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003975666873267118,
      "loss": 1.8136,
      "step": 4230
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039755519222772144,
      "loss": 1.8227,
      "step": 4240
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003975436702079136,
      "loss": 1.7746,
      "step": 4250
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039753212126885843,
      "loss": 1.856,
      "step": 4260
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039752054541212974,
      "loss": 1.8853,
      "step": 4270
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003975089426393049,
      "loss": 1.8294,
      "step": 4280
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039749731295196503,
      "loss": 1.7912,
      "step": 4290
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.000397485656351695,
      "loss": 1.791,
      "step": 4300
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039747397284008303,
      "loss": 1.7691,
      "step": 4310
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003974622624187215,
      "loss": 1.8341,
      "step": 4320
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.000397450525089206,
      "loss": 1.7926,
      "step": 4330
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.000397438760853136,
      "loss": 1.7929,
      "step": 4340
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003974269697121147,
      "loss": 1.8295,
      "step": 4350
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039741515166774876,
      "loss": 1.7966,
      "step": 4360
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039740330672164873,
      "loss": 1.785,
      "step": 4370
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039739143487542867,
      "loss": 1.8314,
      "step": 4380
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039737953613070635,
      "loss": 2.1719,
      "step": 4390
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003973676104891032,
      "loss": 2.1545,
      "step": 4400
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003973556579522444,
      "loss": 1.8765,
      "step": 4410
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039734367852175867,
      "loss": 1.8129,
      "step": 4420
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039733167219927834,
      "loss": 1.8307,
      "step": 4430
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003973196389864397,
      "loss": 1.8195,
      "step": 4440
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039730757888488237,
      "loss": 1.8062,
      "step": 4450
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003972954918962499,
      "loss": 1.8539,
      "step": 4460
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039728337802218925,
      "loss": 1.815,
      "step": 4470
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003972712372643512,
      "loss": 1.7471,
      "step": 4480
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003972590696243903,
      "loss": 1.8931,
      "step": 4490
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003972468751039645,
      "loss": 1.8712,
      "step": 4500
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003972346537047356,
      "loss": 1.7898,
      "step": 4510
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.000397222405428369,
      "loss": 1.8339,
      "step": 4520
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039721013027653376,
      "loss": 1.7983,
      "step": 4530
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003971978282509026,
      "loss": 1.7834,
      "step": 4540
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039718549935315197,
      "loss": 1.8152,
      "step": 4550
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003971731435849619,
      "loss": 1.7459,
      "step": 4560
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039716076094801605,
      "loss": 1.8813,
      "step": 4570
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003971483514440019,
      "loss": 1.8476,
      "step": 4580
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003971359150746104,
      "loss": 1.8125,
      "step": 4590
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039712345184153634,
      "loss": 1.8821,
      "step": 4600
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003971109617464779,
      "loss": 1.8308,
      "step": 4610
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003970984447911374,
      "loss": 1.8794,
      "step": 4620
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039708590097722033,
      "loss": 1.8233,
      "step": 4630
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003970733303064361,
      "loss": 1.8059,
      "step": 4640
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039706073278049755,
      "loss": 1.794,
      "step": 4650
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003970481084011215,
      "loss": 1.788,
      "step": 4660
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039703545717002835,
      "loss": 1.838,
      "step": 4670
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039702277908894193,
      "loss": 1.8613,
      "step": 4680
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003970100741595899,
      "loss": 1.8033,
      "step": 4690
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039699734238370364,
      "loss": 1.8565,
      "step": 4700
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039698458376301803,
      "loss": 1.8227,
      "step": 4710
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003969717982992717,
      "loss": 1.8421,
      "step": 4720
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039695898599420695,
      "loss": 1.7647,
      "step": 4730
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003969461468495697,
      "loss": 1.8001,
      "step": 4740
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039693328086710955,
      "loss": 1.8001,
      "step": 4750
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003969203880485797,
      "loss": 1.782,
      "step": 4760
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.000396907468395737,
      "loss": 1.8169,
      "step": 4770
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003968945219103422,
      "loss": 1.7674,
      "step": 4780
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003968815485941594,
      "loss": 1.8024,
      "step": 4790
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003968685484489564,
      "loss": 1.7552,
      "step": 4800
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003968555214765049,
      "loss": 1.8105,
      "step": 4810
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039684246767857995,
      "loss": 1.8472,
      "step": 4820
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039682938705696045,
      "loss": 1.8428,
      "step": 4830
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003968162796134288,
      "loss": 1.7878,
      "step": 4840
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003968031453497712,
      "loss": 1.773,
      "step": 4850
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039678998426777756,
      "loss": 1.7859,
      "step": 4860
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003967767963692413,
      "loss": 1.8525,
      "step": 4870
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003967635816559594,
      "loss": 1.8496,
      "step": 4880
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003967503401297328,
      "loss": 1.7596,
      "step": 4890
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039673707179236576,
      "loss": 1.8011,
      "step": 4900
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039672377664566643,
      "loss": 1.9062,
      "step": 4910
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003967104546914466,
      "loss": 1.7891,
      "step": 4920
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039669710593152153,
      "loss": 1.7634,
      "step": 4930
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003966837303677103,
      "loss": 1.8714,
      "step": 4940
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003966703280018356,
      "loss": 1.7708,
      "step": 4950
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039665689883572385,
      "loss": 1.7492,
      "step": 4960
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039664344287120495,
      "loss": 1.7867,
      "step": 4970
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003966299601101125,
      "loss": 1.7348,
      "step": 4980
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003966164505542839,
      "loss": 1.7343,
      "step": 4990
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039660291420556007,
      "loss": 1.8217,
      "step": 5000
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003965893510657856,
      "loss": 1.7967,
      "step": 5010
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003965757611368086,
      "loss": 1.8114,
      "step": 5020
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003965621444204812,
      "loss": 1.7517,
      "step": 5030
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003965485009186588,
      "loss": 1.7804,
      "step": 5040
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039653483063320066,
      "loss": 1.8003,
      "step": 5050
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003965211335659696,
      "loss": 1.7153,
      "step": 5060
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003965074097188321,
      "loss": 1.8146,
      "step": 5070
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003964936590936583,
      "loss": 1.7414,
      "step": 5080
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003964798816923221,
      "loss": 1.7686,
      "step": 5090
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039646607751670084,
      "loss": 1.8036,
      "step": 5100
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003964522465686757,
      "loss": 1.8148,
      "step": 5110
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039643838885013123,
      "loss": 1.7598,
      "step": 5120
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.000396424504362956,
      "loss": 1.7619,
      "step": 5130
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.000396410593109042,
      "loss": 1.8385,
      "step": 5140
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.000396396655090285,
      "loss": 1.8058,
      "step": 5150
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039638269030858414,
      "loss": 1.7824,
      "step": 5160
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003963686987658426,
      "loss": 1.7732,
      "step": 5170
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003963546804639668,
      "loss": 2.0963,
      "step": 5180
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039634063540486714,
      "loss": 1.9065,
      "step": 5190
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003963265635904576,
      "loss": 1.7381,
      "step": 5200
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003963124650226556,
      "loss": 1.8123,
      "step": 5210
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003962983397033824,
      "loss": 1.7298,
      "step": 5220
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039628418763456295,
      "loss": 1.7675,
      "step": 5230
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039627000881812554,
      "loss": 1.7765,
      "step": 5240
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039625580325600255,
      "loss": 1.7708,
      "step": 5250
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003962415709501296,
      "loss": 1.8024,
      "step": 5260
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003962273119024462,
      "loss": 1.8149,
      "step": 5270
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003962130261148954,
      "loss": 1.7905,
      "step": 5280
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039619871358942405,
      "loss": 1.7407,
      "step": 5290
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039618437432798235,
      "loss": 1.7789,
      "step": 5300
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039617000833252434,
      "loss": 1.74,
      "step": 5310
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003961556156050077,
      "loss": 1.7821,
      "step": 5320
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003961411961473937,
      "loss": 1.8277,
      "step": 5330
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003961267499616473,
      "loss": 1.7875,
      "step": 5340
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003961122770497371,
      "loss": 1.7887,
      "step": 5350
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003960977774136353,
      "loss": 1.7135,
      "step": 5360
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039608325105531774,
      "loss": 1.7723,
      "step": 5370
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.000396068697976764,
      "loss": 1.7617,
      "step": 5380
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003960541181799571,
      "loss": 1.7773,
      "step": 5390
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.000396039511666884,
      "loss": 1.7796,
      "step": 5400
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039602487843953495,
      "loss": 1.8129,
      "step": 5410
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003960102184999041,
      "loss": 1.7779,
      "step": 5420
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003959955318499892,
      "loss": 1.8043,
      "step": 5430
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003959808184917915,
      "loss": 1.8121,
      "step": 5440
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039596607842731606,
      "loss": 1.8458,
      "step": 5450
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039595131165857147,
      "loss": 1.6925,
      "step": 5460
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039593651818757005,
      "loss": 1.7781,
      "step": 5470
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003959216980163276,
      "loss": 1.7734,
      "step": 5480
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003959068511468638,
      "loss": 1.7421,
      "step": 5490
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003958919775812017,
      "loss": 1.7665,
      "step": 5500
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003958770773213682,
      "loss": 1.8133,
      "step": 5510
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003958621503693937,
      "loss": 1.8489,
      "step": 5520
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039584719672731234,
      "loss": 1.7686,
      "step": 5530
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039583221639716184,
      "loss": 1.7774,
      "step": 5540
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003958172093809836,
      "loss": 1.7615,
      "step": 5550
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039580217568082254,
      "loss": 1.7265,
      "step": 5560
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003957871152987274,
      "loss": 1.7839,
      "step": 5570
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003957720282367504,
      "loss": 1.8069,
      "step": 5580
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039575691449694744,
      "loss": 1.7924,
      "step": 5590
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039574177408137806,
      "loss": 1.7417,
      "step": 5600
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039572660699210545,
      "loss": 1.7806,
      "step": 5610
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003957114132311965,
      "loss": 1.7968,
      "step": 5620
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003956961928007216,
      "loss": 1.7929,
      "step": 5630
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003956809457027548,
      "loss": 1.7942,
      "step": 5640
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003956656719393739,
      "loss": 1.7724,
      "step": 5650
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039565037151266015,
      "loss": 1.7645,
      "step": 5660
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039563504442469865,
      "loss": 1.825,
      "step": 5670
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039561969067757795,
      "loss": 1.7706,
      "step": 5680
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039560431027339034,
      "loss": 1.6977,
      "step": 5690
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039558890321423165,
      "loss": 1.821,
      "step": 5700
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039557346950220147,
      "loss": 1.8348,
      "step": 5710
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039555800913940295,
      "loss": 1.7382,
      "step": 5720
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003955425221279428,
      "loss": 1.8237,
      "step": 5730
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039552700846993147,
      "loss": 1.7565,
      "step": 5740
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.000395511468167483,
      "loss": 1.755,
      "step": 5750
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039549590122271505,
      "loss": 1.7847,
      "step": 5760
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.000395480307637749,
      "loss": 1.7783,
      "step": 5770
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003954646874147097,
      "loss": 1.794,
      "step": 5780
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003954490405557257,
      "loss": 1.8117,
      "step": 5790
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003954333670629293,
      "loss": 1.7388,
      "step": 5800
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039541766693845624,
      "loss": 1.7699,
      "step": 5810
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039540194018444606,
      "loss": 1.7813,
      "step": 5820
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039538618680304174,
      "loss": 1.7695,
      "step": 5830
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039537040679638995,
      "loss": 1.7895,
      "step": 5840
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003953546001666412,
      "loss": 1.7183,
      "step": 5850
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003953387669159494,
      "loss": 1.7388,
      "step": 5860
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003953229070464721,
      "loss": 1.7812,
      "step": 5870
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003953070205603704,
      "loss": 1.7602,
      "step": 5880
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003952911074598094,
      "loss": 1.7441,
      "step": 5890
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003952751677469575,
      "loss": 1.7681,
      "step": 5900
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039525920142398667,
      "loss": 1.7261,
      "step": 5910
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003952432084930728,
      "loss": 1.7352,
      "step": 5920
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039522718895639513,
      "loss": 1.7927,
      "step": 5930
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039521114281613676,
      "loss": 1.7422,
      "step": 5940
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003951950700744841,
      "loss": 1.8138,
      "step": 5950
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003951789707336276,
      "loss": 1.7728,
      "step": 5960
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039516284479576097,
      "loss": 1.7646,
      "step": 5970
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039514669226308174,
      "loss": 1.8766,
      "step": 5980
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039513051313779104,
      "loss": 1.7573,
      "step": 5990
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039511430742209354,
      "loss": 1.774,
      "step": 6000
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003950980751181976,
      "loss": 1.7494,
      "step": 6010
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003950818162283152,
      "loss": 1.8108,
      "step": 6020
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039506553075466195,
      "loss": 1.7655,
      "step": 6030
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039504921869945716,
      "loss": 1.735,
      "step": 6040
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003950328800649235,
      "loss": 1.8083,
      "step": 6050
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039501651485328754,
      "loss": 1.7709,
      "step": 6060
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039500012306677927,
      "loss": 1.7238,
      "step": 6070
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003949837047076325,
      "loss": 1.7706,
      "step": 6080
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003949672597780845,
      "loss": 1.731,
      "step": 6090
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003949507882803763,
      "loss": 1.7088,
      "step": 6100
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003949342902167524,
      "loss": 1.797,
      "step": 6110
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003949177655894609,
      "loss": 1.7863,
      "step": 6120
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003949012144007538,
      "loss": 1.8108,
      "step": 6130
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003948846366528864,
      "loss": 1.8412,
      "step": 6140
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039486803234811777,
      "loss": 1.8077,
      "step": 6150
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003948514014887106,
      "loss": 1.7675,
      "step": 6160
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003948347440769311,
      "loss": 1.8441,
      "step": 6170
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003948180601150493,
      "loss": 1.7729,
      "step": 6180
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003948013496053387,
      "loss": 1.7453,
      "step": 6190
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003947846125500763,
      "loss": 1.7238,
      "step": 6200
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039476784895154293,
      "loss": 1.7232,
      "step": 6210
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039475105881202306,
      "loss": 1.7615,
      "step": 6220
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039473424213380454,
      "loss": 1.7313,
      "step": 6230
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003947173989191791,
      "loss": 1.7711,
      "step": 6240
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039470052917044184,
      "loss": 1.754,
      "step": 6250
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003946836328898917,
      "loss": 1.7765,
      "step": 6260
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003946667100798311,
      "loss": 1.7149,
      "step": 6270
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039464976074256606,
      "loss": 1.7595,
      "step": 6280
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039463278488040634,
      "loss": 1.7473,
      "step": 6290
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003946157824956652,
      "loss": 1.6789,
      "step": 6300
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039459875359065957,
      "loss": 1.7578,
      "step": 6310
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039458169816770987,
      "loss": 1.7213,
      "step": 6320
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039456461622914044,
      "loss": 1.7212,
      "step": 6330
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039454750777727886,
      "loss": 1.7128,
      "step": 6340
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039453037281445664,
      "loss": 1.7406,
      "step": 6350
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003945132113430086,
      "loss": 1.8279,
      "step": 6360
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003944960233652734,
      "loss": 1.7301,
      "step": 6370
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003944788088835933,
      "loss": 1.7591,
      "step": 6380
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003944615679003141,
      "loss": 1.769,
      "step": 6390
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039444430041778514,
      "loss": 1.6983,
      "step": 6400
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003944270064383595,
      "loss": 1.7071,
      "step": 6410
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039440968596439393,
      "loss": 1.7247,
      "step": 6420
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039439233899824855,
      "loss": 1.7116,
      "step": 6430
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039437496554228725,
      "loss": 1.7587,
      "step": 6440
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039435756559887754,
      "loss": 1.7289,
      "step": 6450
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039434013917039057,
      "loss": 1.7415,
      "step": 6460
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003943226862592009,
      "loss": 1.8086,
      "step": 6470
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039430520686768695,
      "loss": 1.8101,
      "step": 6480
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039428770099823055,
      "loss": 1.7399,
      "step": 6490
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003942701686532173,
      "loss": 1.7788,
      "step": 6500
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039425260983503625,
      "loss": 1.7449,
      "step": 6510
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003942350245460802,
      "loss": 1.7242,
      "step": 6520
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039421741278874555,
      "loss": 1.7774,
      "step": 6530
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003941997745654321,
      "loss": 1.6937,
      "step": 6540
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003941821098785435,
      "loss": 1.7575,
      "step": 6550
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.000394164418730487,
      "loss": 1.7652,
      "step": 6560
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003941467011236731,
      "loss": 1.7747,
      "step": 6570
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039412895706051644,
      "loss": 1.7496,
      "step": 6580
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003941111865434349,
      "loss": 1.7113,
      "step": 6590
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039409338957485006,
      "loss": 1.7558,
      "step": 6600
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039407556615718713,
      "loss": 1.7222,
      "step": 6610
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003940577162928749,
      "loss": 1.7327,
      "step": 6620
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039403983998434575,
      "loss": 1.7117,
      "step": 6630
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039402193723403576,
      "loss": 1.7442,
      "step": 6640
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003940040080443844,
      "loss": 1.708,
      "step": 6650
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039398605241783497,
      "loss": 1.7873,
      "step": 6660
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039396807035683426,
      "loss": 1.7372,
      "step": 6670
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003939500618638327,
      "loss": 1.7293,
      "step": 6680
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039393202694128427,
      "loss": 1.7413,
      "step": 6690
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003939139655916466,
      "loss": 1.7382,
      "step": 6700
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039389587781738097,
      "loss": 1.7376,
      "step": 6710
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003938777636209521,
      "loss": 1.8539,
      "step": 6720
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003938596230048285,
      "loss": 2.0847,
      "step": 6730
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039384145597148206,
      "loss": 1.7983,
      "step": 6740
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039382326252338857,
      "loss": 1.7416,
      "step": 6750
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039380504266302703,
      "loss": 1.7959,
      "step": 6760
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039378679639288054,
      "loss": 1.7394,
      "step": 6770
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039376852371543524,
      "loss": 1.7364,
      "step": 6780
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003937502246331813,
      "loss": 1.7741,
      "step": 6790
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003937318991486123,
      "loss": 1.7287,
      "step": 6800
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003937135472642255,
      "loss": 1.6729,
      "step": 6810
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003936951689825216,
      "loss": 1.7514,
      "step": 6820
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039367676430600507,
      "loss": 1.7261,
      "step": 6830
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003936583332371839,
      "loss": 1.6996,
      "step": 6840
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039363987577856977,
      "loss": 1.7291,
      "step": 6850
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003936213919326777,
      "loss": 1.7375,
      "step": 6860
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003936028817020267,
      "loss": 1.773,
      "step": 6870
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039358434508913895,
      "loss": 1.7226,
      "step": 6880
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039356578209654054,
      "loss": 1.7358,
      "step": 6890
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039354719272676104,
      "loss": 1.7529,
      "step": 6900
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039352857698233363,
      "loss": 1.7016,
      "step": 6910
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039350993486579497,
      "loss": 1.7673,
      "step": 6920
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039349126637968566,
      "loss": 1.8237,
      "step": 6930
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039347257152654936,
      "loss": 1.7175,
      "step": 6940
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039345385030893377,
      "loss": 1.7287,
      "step": 6950
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039343510272939,
      "loss": 1.7156,
      "step": 6960
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003934163287904729,
      "loss": 1.6863,
      "step": 6970
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039339752849474057,
      "loss": 1.7854,
      "step": 6980
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.000393378701844755,
      "loss": 1.7371,
      "step": 6990
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003933598488430818,
      "loss": 1.7483,
      "step": 7000
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039334096949228994,
      "loss": 1.7263,
      "step": 7010
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003933220637949522,
      "loss": 1.7644,
      "step": 7020
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0003933031317536448,
      "loss": 1.752,
      "step": 7030
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.15631213784217834,
      "learning_rate": 0.00039328417337094763,
      "loss": 1.7599,
      "step": 7040
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.14645032584667206,
      "learning_rate": 0.00039326518864944417,
      "loss": 1.7166,
      "step": 7050
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1466301679611206,
      "learning_rate": 0.00039324617759172137,
      "loss": 1.7425,
      "step": 7060
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.13789162039756775,
      "learning_rate": 0.00039322714020037,
      "loss": 1.7232,
      "step": 7070
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1505664587020874,
      "learning_rate": 0.00039320807647798414,
      "loss": 1.6992,
      "step": 7080
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.131247878074646,
      "learning_rate": 0.00039318898642716173,
      "loss": 1.7142,
      "step": 7090
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.13916432857513428,
      "learning_rate": 0.0003931698700505041,
      "loss": 1.7512,
      "step": 7100
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.13484828174114227,
      "learning_rate": 0.0003931507273506162,
      "loss": 1.7542,
      "step": 7110
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.14107249677181244,
      "learning_rate": 0.00039313155833010666,
      "loss": 1.6634,
      "step": 7120
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.14845772087574005,
      "learning_rate": 0.0003931123629915876,
      "loss": 1.7984,
      "step": 7130
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1481531858444214,
      "learning_rate": 0.0003930931413376748,
      "loss": 1.781,
      "step": 7140
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.14172106981277466,
      "learning_rate": 0.00039307389337098767,
      "loss": 1.7845,
      "step": 7150
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.14321057498455048,
      "learning_rate": 0.000393054619094149,
      "loss": 1.7123,
      "step": 7160
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.14234977960586548,
      "learning_rate": 0.0003930353185097853,
      "loss": 1.7058,
      "step": 7170
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.17808882892131805,
      "learning_rate": 0.00039301599162052677,
      "loss": 1.7017,
      "step": 7180
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.13419701159000397,
      "learning_rate": 0.000392996638429007,
      "loss": 1.7307,
      "step": 7190
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1409374326467514,
      "learning_rate": 0.00039297725893786315,
      "loss": 1.7113,
      "step": 7200
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.13434813916683197,
      "learning_rate": 0.0003929578531497362,
      "loss": 1.6991,
      "step": 7210
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.13483397662639618,
      "learning_rate": 0.0003929384210672706,
      "loss": 1.7135,
      "step": 7220
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.15328499674797058,
      "learning_rate": 0.00039291896269311427,
      "loss": 1.7431,
      "step": 7230
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.14323565363883972,
      "learning_rate": 0.0003928994780299188,
      "loss": 1.7087,
      "step": 7240
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.14380577206611633,
      "learning_rate": 0.0003928799670803394,
      "loss": 1.7823,
      "step": 7250
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.13544294238090515,
      "learning_rate": 0.00039286042984703484,
      "loss": 1.7303,
      "step": 7260
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.13900905847549438,
      "learning_rate": 0.0003928408663326674,
      "loss": 1.7391,
      "step": 7270
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1786203235387802,
      "learning_rate": 0.00039282127653990297,
      "loss": 1.7404,
      "step": 7280
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.15546490252017975,
      "learning_rate": 0.0003928016604714112,
      "loss": 1.7522,
      "step": 7290
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.16079726815223694,
      "learning_rate": 0.000392782018129865,
      "loss": 1.7286,
      "step": 7300
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.14534011483192444,
      "learning_rate": 0.0003927623495179411,
      "loss": 1.7904,
      "step": 7310
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.14322422444820404,
      "learning_rate": 0.0003927426546383198,
      "loss": 1.6674,
      "step": 7320
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.139568492770195,
      "learning_rate": 0.00039272293349368483,
      "loss": 1.7203,
      "step": 7330
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.21763065457344055,
      "learning_rate": 0.0003927031860867236,
      "loss": 1.731,
      "step": 7340
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.13749991357326508,
      "learning_rate": 0.00039268341242012714,
      "loss": 1.709,
      "step": 7350
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.14703145623207092,
      "learning_rate": 0.0003926636124965899,
      "loss": 1.6717,
      "step": 7360
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1383516639471054,
      "learning_rate": 0.00039264378631881017,
      "loss": 1.6542,
      "step": 7370
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1362847089767456,
      "learning_rate": 0.0003926239338894895,
      "loss": 1.7087,
      "step": 7380
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1404048204421997,
      "learning_rate": 0.00039260405521133324,
      "loss": 1.7072,
      "step": 7390
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.16335636377334595,
      "learning_rate": 0.0003925841502870503,
      "loss": 1.7198,
      "step": 7400
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.13588540256023407,
      "learning_rate": 0.00039256421911935306,
      "loss": 1.6984,
      "step": 7410
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.15666501224040985,
      "learning_rate": 0.0003925442617109575,
      "loss": 1.7881,
      "step": 7420
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.14240530133247375,
      "learning_rate": 0.00039252427806458346,
      "loss": 1.698,
      "step": 7430
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.14285792410373688,
      "learning_rate": 0.0003925042681829537,
      "loss": 1.7634,
      "step": 7440
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.17508766055107117,
      "learning_rate": 0.0003924842320687953,
      "loss": 1.6889,
      "step": 7450
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.13306574523448944,
      "learning_rate": 0.0003924641697248385,
      "loss": 1.6307,
      "step": 7460
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1410142034292221,
      "learning_rate": 0.0003924440811538171,
      "loss": 1.7104,
      "step": 7470
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.14873316884040833,
      "learning_rate": 0.0003924239663584686,
      "loss": 1.7204,
      "step": 7480
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.13496127724647522,
      "learning_rate": 0.0003924038253415341,
      "loss": 1.6493,
      "step": 7490
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1559535413980484,
      "learning_rate": 0.00039238365810575815,
      "loss": 1.7679,
      "step": 7500
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1314055621623993,
      "learning_rate": 0.000392363464653889,
      "loss": 1.6889,
      "step": 7510
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.13722346723079681,
      "learning_rate": 0.00039234324498867835,
      "loss": 1.7167,
      "step": 7520
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.17334945499897003,
      "learning_rate": 0.0003923229991128816,
      "loss": 1.7667,
      "step": 7530
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.13046298921108246,
      "learning_rate": 0.00039230272702925757,
      "loss": 1.6632,
      "step": 7540
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.13783209025859833,
      "learning_rate": 0.00039228242874056875,
      "loss": 1.7177,
      "step": 7550
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.16211003065109253,
      "learning_rate": 0.0003922621042495812,
      "loss": 1.7712,
      "step": 7560
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1397620439529419,
      "learning_rate": 0.0003922417535590646,
      "loss": 1.6844,
      "step": 7570
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1390129178762436,
      "learning_rate": 0.00039222137667179195,
      "loss": 1.6965,
      "step": 7580
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.14877186715602875,
      "learning_rate": 0.00039220097359054027,
      "loss": 1.6844,
      "step": 7590
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.13841237127780914,
      "learning_rate": 0.0003921805443180897,
      "loss": 1.6794,
      "step": 7600
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.14043334126472473,
      "learning_rate": 0.00039216008885722416,
      "loss": 1.7271,
      "step": 7610
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2009606659412384,
      "learning_rate": 0.0003921396072107311,
      "loss": 1.8196,
      "step": 7620
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.170047327876091,
      "learning_rate": 0.0003921190993814016,
      "loss": 1.7412,
      "step": 7630
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.16911502182483673,
      "learning_rate": 0.0003920985653720303,
      "loss": 1.7761,
      "step": 7640
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.15350545942783356,
      "learning_rate": 0.00039207800518541527,
      "loss": 1.7545,
      "step": 7650
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.14801712334156036,
      "learning_rate": 0.0003920574188243583,
      "loss": 1.6783,
      "step": 7660
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1496005654335022,
      "learning_rate": 0.0003920368062916647,
      "loss": 1.6807,
      "step": 7670
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.14550770819187164,
      "learning_rate": 0.00039201616759014323,
      "loss": 1.7223,
      "step": 7680
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.14356763660907745,
      "learning_rate": 0.00039199550272260645,
      "loss": 1.7058,
      "step": 7690
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.14512434601783752,
      "learning_rate": 0.0003919748116918703,
      "loss": 1.7302,
      "step": 7700
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1333114504814148,
      "learning_rate": 0.00039195409450075436,
      "loss": 1.7544,
      "step": 7710
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.14479930698871613,
      "learning_rate": 0.00039193335115208177,
      "loss": 1.6746,
      "step": 7720
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.13208623230457306,
      "learning_rate": 0.00039191258164867916,
      "loss": 1.6208,
      "step": 7730
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1354825794696808,
      "learning_rate": 0.0003918917859933769,
      "loss": 1.6876,
      "step": 7740
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.13403142988681793,
      "learning_rate": 0.0003918709641890087,
      "loss": 1.7227,
      "step": 7750
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.14381615817546844,
      "learning_rate": 0.000391850116238412,
      "loss": 1.7347,
      "step": 7760
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.148904949426651,
      "learning_rate": 0.0003918292421444277,
      "loss": 1.7538,
      "step": 7770
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.14611361920833588,
      "learning_rate": 0.0003918083419099004,
      "loss": 1.7352,
      "step": 7780
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.13526111841201782,
      "learning_rate": 0.0003917874155376781,
      "loss": 1.7334,
      "step": 7790
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.14207859337329865,
      "learning_rate": 0.00039176646303061243,
      "loss": 1.7621,
      "step": 7800
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.15307915210723877,
      "learning_rate": 0.00039174548439155864,
      "loss": 1.7134,
      "step": 7810
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.14540807902812958,
      "learning_rate": 0.0003917244796233754,
      "loss": 1.7038,
      "step": 7820
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.14969046413898468,
      "learning_rate": 0.00039170344872892514,
      "loss": 1.6982,
      "step": 7830
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1437288224697113,
      "learning_rate": 0.0003916823917110737,
      "loss": 1.7362,
      "step": 7840
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.13481366634368896,
      "learning_rate": 0.00039166130857269037,
      "loss": 1.757,
      "step": 7850
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.16136403381824493,
      "learning_rate": 0.0003916401993166484,
      "loss": 1.6612,
      "step": 7860
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.13674139976501465,
      "learning_rate": 0.0003916190639458242,
      "loss": 1.6701,
      "step": 7870
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.155400812625885,
      "learning_rate": 0.0003915979024630978,
      "loss": 1.7115,
      "step": 7880
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.15934404730796814,
      "learning_rate": 0.000391576714871353,
      "loss": 1.7014,
      "step": 7890
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.13780388236045837,
      "learning_rate": 0.00039155550117347704,
      "loss": 1.6889,
      "step": 7900
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.14378343522548676,
      "learning_rate": 0.0003915342613723607,
      "loss": 1.7577,
      "step": 7910
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2770020067691803,
      "learning_rate": 0.00039151299547089826,
      "loss": 1.7168,
      "step": 7920
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.16923096776008606,
      "learning_rate": 0.0003914917034719876,
      "loss": 1.7309,
      "step": 7930
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.14824189245700836,
      "learning_rate": 0.0003914703853785303,
      "loss": 1.7077,
      "step": 7940
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.148031547665596,
      "learning_rate": 0.0003914490411934313,
      "loss": 1.7435,
      "step": 7950
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.14204919338226318,
      "learning_rate": 0.0003914276709195991,
      "loss": 1.7234,
      "step": 7960
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.15563704073429108,
      "learning_rate": 0.000391406274559946,
      "loss": 1.7848,
      "step": 7970
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.14522221684455872,
      "learning_rate": 0.00039138485211738755,
      "loss": 1.7095,
      "step": 7980
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.14838719367980957,
      "learning_rate": 0.00039136340359484297,
      "loss": 1.726,
      "step": 7990
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.15066173672676086,
      "learning_rate": 0.00039134192899523514,
      "loss": 1.7538,
      "step": 8000
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.15595562756061554,
      "learning_rate": 0.00039132042832149027,
      "loss": 1.6987,
      "step": 8010
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1424400955438614,
      "learning_rate": 0.0003912989015765384,
      "loss": 1.6906,
      "step": 8020
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.14538030326366425,
      "learning_rate": 0.0003912773487633129,
      "loss": 1.6979,
      "step": 8030
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1990845650434494,
      "learning_rate": 0.00039125576988475075,
      "loss": 1.7169,
      "step": 8040
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.14819040894508362,
      "learning_rate": 0.0003912341649437925,
      "loss": 1.6725,
      "step": 8050
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.23003418743610382,
      "learning_rate": 0.0003912125339433824,
      "loss": 1.7566,
      "step": 8060
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1365618258714676,
      "learning_rate": 0.0003911908768864679,
      "loss": 1.7369,
      "step": 8070
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1465931236743927,
      "learning_rate": 0.0003911691937760003,
      "loss": 1.7282,
      "step": 8080
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.13349132239818573,
      "learning_rate": 0.0003911474846149344,
      "loss": 1.6716,
      "step": 8090
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1357334852218628,
      "learning_rate": 0.00039112574940622835,
      "loss": 1.7131,
      "step": 8100
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.14389128983020782,
      "learning_rate": 0.0003911039881528441,
      "loss": 1.7469,
      "step": 8110
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.13784164190292358,
      "learning_rate": 0.0003910822008577471,
      "loss": 1.7734,
      "step": 8120
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.13456936180591583,
      "learning_rate": 0.00039106038752390623,
      "loss": 1.6929,
      "step": 8130
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.12721242010593414,
      "learning_rate": 0.00039103854815429406,
      "loss": 1.6928,
      "step": 8140
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.12918342649936676,
      "learning_rate": 0.00039101668275188655,
      "loss": 1.7344,
      "step": 8150
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1323053538799286,
      "learning_rate": 0.0003909947913196633,
      "loss": 1.6671,
      "step": 8160
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.13387516140937805,
      "learning_rate": 0.00039097287386060753,
      "loss": 1.657,
      "step": 8170
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1323385089635849,
      "learning_rate": 0.0003909509303777059,
      "loss": 1.6923,
      "step": 8180
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.12993253767490387,
      "learning_rate": 0.0003909289608739486,
      "loss": 1.6682,
      "step": 8190
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.14957071840763092,
      "learning_rate": 0.0003909069653523295,
      "loss": 1.6946,
      "step": 8200
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1388004869222641,
      "learning_rate": 0.0003908849438158458,
      "loss": 1.7008,
      "step": 8210
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1595703661441803,
      "learning_rate": 0.00039086289626749854,
      "loss": 1.7514,
      "step": 8220
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1537179946899414,
      "learning_rate": 0.0003908408227102919,
      "loss": 1.695,
      "step": 8230
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1576649695634842,
      "learning_rate": 0.0003908187231472341,
      "loss": 1.6994,
      "step": 8240
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.14125628769397736,
      "learning_rate": 0.0003907965975813365,
      "loss": 1.7204,
      "step": 8250
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1359732747077942,
      "learning_rate": 0.0003907744460156142,
      "loss": 1.7711,
      "step": 8260
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1417861431837082,
      "learning_rate": 0.0003907522684530857,
      "loss": 1.7593,
      "step": 8270
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.16784122586250305,
      "learning_rate": 0.00039073006489677317,
      "loss": 1.7329,
      "step": 8280
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.14502134919166565,
      "learning_rate": 0.00039070783534970235,
      "loss": 1.6954,
      "step": 8290
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13937999308109283,
      "learning_rate": 0.00039068557981490246,
      "loss": 1.6908,
      "step": 8300
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1434570550918579,
      "learning_rate": 0.00039066329829540614,
      "loss": 1.7373,
      "step": 8310
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13864974677562714,
      "learning_rate": 0.00039064099079424985,
      "loss": 1.7309,
      "step": 8320
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.17600135505199432,
      "learning_rate": 0.00039061865731447327,
      "loss": 1.6646,
      "step": 8330
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.14923590421676636,
      "learning_rate": 0.00039059629785911985,
      "loss": 1.7094,
      "step": 8340
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.16249333322048187,
      "learning_rate": 0.00039057391243123654,
      "loss": 1.672,
      "step": 8350
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.15163445472717285,
      "learning_rate": 0.0003905515010338738,
      "loss": 1.7534,
      "step": 8360
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13050870597362518,
      "learning_rate": 0.00039052906367008556,
      "loss": 1.6948,
      "step": 8370
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.14099529385566711,
      "learning_rate": 0.0003905066003429294,
      "loss": 1.7127,
      "step": 8380
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13801635801792145,
      "learning_rate": 0.00039048411105546644,
      "loss": 1.7082,
      "step": 8390
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13237683475017548,
      "learning_rate": 0.00039046159581076123,
      "loss": 1.6959,
      "step": 8400
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.144191712141037,
      "learning_rate": 0.00039043905461188193,
      "loss": 1.6596,
      "step": 8410
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1366361379623413,
      "learning_rate": 0.0003904164874619002,
      "loss": 1.7135,
      "step": 8420
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13347311317920685,
      "learning_rate": 0.0003903938943638913,
      "loss": 1.6979,
      "step": 8430
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1298627257347107,
      "learning_rate": 0.000390371275320934,
      "loss": 1.6883,
      "step": 8440
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.15101325511932373,
      "learning_rate": 0.0003903486303361106,
      "loss": 1.7202,
      "step": 8450
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13881361484527588,
      "learning_rate": 0.00039032595941250687,
      "loss": 1.7568,
      "step": 8460
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1474679857492447,
      "learning_rate": 0.0003903032625532122,
      "loss": 1.7472,
      "step": 8470
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13569600880146027,
      "learning_rate": 0.00039028053976131947,
      "loss": 1.6686,
      "step": 8480
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.15183793008327484,
      "learning_rate": 0.0003902577910399252,
      "loss": 1.7271,
      "step": 8490
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.14885376393795013,
      "learning_rate": 0.00039023501639212925,
      "loss": 1.7249,
      "step": 8500
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13697759807109833,
      "learning_rate": 0.0003902122158210352,
      "loss": 1.7183,
      "step": 8510
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13530884683132172,
      "learning_rate": 0.00039018938932974996,
      "loss": 1.7494,
      "step": 8520
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13527202606201172,
      "learning_rate": 0.0003901665369213842,
      "loss": 1.7127,
      "step": 8530
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13087345659732819,
      "learning_rate": 0.000390143658599052,
      "loss": 1.7008,
      "step": 8540
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1484183967113495,
      "learning_rate": 0.00039012075436587097,
      "loss": 1.719,
      "step": 8550
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13455341756343842,
      "learning_rate": 0.0003900978242249623,
      "loss": 1.6938,
      "step": 8560
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13760384917259216,
      "learning_rate": 0.0003900748681794506,
      "loss": 1.685,
      "step": 8570
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13497385382652283,
      "learning_rate": 0.00039005188623246413,
      "loss": 1.7293,
      "step": 8580
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1361473649740219,
      "learning_rate": 0.0003900288783871347,
      "loss": 1.7187,
      "step": 8590
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13627833127975464,
      "learning_rate": 0.00039000584464659753,
      "loss": 1.6916,
      "step": 8600
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13532902300357819,
      "learning_rate": 0.00038998278501399135,
      "loss": 1.747,
      "step": 8610
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13195374608039856,
      "learning_rate": 0.00038995969949245853,
      "loss": 1.6929,
      "step": 8620
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13569962978363037,
      "learning_rate": 0.000389936588085145,
      "loss": 1.65,
      "step": 8630
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.15099965035915375,
      "learning_rate": 0.00038991345079520014,
      "loss": 1.7891,
      "step": 8640
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13594239950180054,
      "learning_rate": 0.0003898902876257768,
      "loss": 1.6748,
      "step": 8650
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.14765936136245728,
      "learning_rate": 0.0003898670985800315,
      "loss": 1.7528,
      "step": 8660
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13044315576553345,
      "learning_rate": 0.00038984388366112416,
      "loss": 1.6972,
      "step": 8670
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13899655640125275,
      "learning_rate": 0.0003898206428722182,
      "loss": 1.6653,
      "step": 8680
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13996469974517822,
      "learning_rate": 0.0003897973762164808,
      "loss": 1.6601,
      "step": 8690
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.14995421469211578,
      "learning_rate": 0.0003897740836970825,
      "loss": 1.7096,
      "step": 8700
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.14182275533676147,
      "learning_rate": 0.00038975076531719716,
      "loss": 1.71,
      "step": 8710
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.14055702090263367,
      "learning_rate": 0.0003897274210800025,
      "loss": 1.721,
      "step": 8720
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.14095278084278107,
      "learning_rate": 0.0003897040509886797,
      "loss": 1.6573,
      "step": 8730
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1559416651725769,
      "learning_rate": 0.00038968065504641327,
      "loss": 1.6651,
      "step": 8740
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.14234113693237305,
      "learning_rate": 0.00038965723325639154,
      "loss": 1.7532,
      "step": 8750
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1320653259754181,
      "learning_rate": 0.000389633785621806,
      "loss": 1.704,
      "step": 8760
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.14077289402484894,
      "learning_rate": 0.000389610312145852,
      "loss": 1.6925,
      "step": 8770
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.15022748708724976,
      "learning_rate": 0.0003895868128317282,
      "loss": 1.6837,
      "step": 8780
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.14192801713943481,
      "learning_rate": 0.0003895632876826369,
      "loss": 1.6562,
      "step": 8790
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1271694451570511,
      "learning_rate": 0.0003895397367017838,
      "loss": 1.6415,
      "step": 8800
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.14491376280784607,
      "learning_rate": 0.00038951615989237816,
      "loss": 1.6486,
      "step": 8810
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13827703893184662,
      "learning_rate": 0.0003894925572576329,
      "loss": 1.6637,
      "step": 8820
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13990582525730133,
      "learning_rate": 0.0003894689288007643,
      "loss": 1.7087,
      "step": 8830
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.14701290428638458,
      "learning_rate": 0.0003894452745249922,
      "loss": 1.651,
      "step": 8840
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.21782691776752472,
      "learning_rate": 0.00038942159443353997,
      "loss": 1.6388,
      "step": 8850
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.16319188475608826,
      "learning_rate": 0.0003893978885296345,
      "loss": 1.6662,
      "step": 8860
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.14546512067317963,
      "learning_rate": 0.0003893741568165062,
      "loss": 1.6514,
      "step": 8870
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.145062655210495,
      "learning_rate": 0.0003893503992973889,
      "loss": 1.6861,
      "step": 8880
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.14725039899349213,
      "learning_rate": 0.00038932661597552016,
      "loss": 1.7311,
      "step": 8890
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.15224291384220123,
      "learning_rate": 0.00038930280685414085,
      "loss": 1.7714,
      "step": 8900
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.14499187469482422,
      "learning_rate": 0.0003892789719364955,
      "loss": 1.6955,
      "step": 8910
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13085386157035828,
      "learning_rate": 0.00038925511122583196,
      "loss": 1.649,
      "step": 8920
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.12321346998214722,
      "learning_rate": 0.0003892312247254019,
      "loss": 1.6696,
      "step": 8930
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1457478255033493,
      "learning_rate": 0.00038920731243846017,
      "loss": 1.6824,
      "step": 8940
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1479731947183609,
      "learning_rate": 0.0003891833743682654,
      "loss": 1.6955,
      "step": 8950
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.14455614984035492,
      "learning_rate": 0.00038915941051807963,
      "loss": 1.7087,
      "step": 8960
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1325332075357437,
      "learning_rate": 0.00038913542089116836,
      "loss": 1.6943,
      "step": 8970
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.14660602807998657,
      "learning_rate": 0.00038911140549080064,
      "loss": 1.716,
      "step": 8980
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1645248383283615,
      "learning_rate": 0.0003890873643202491,
      "loss": 1.7111,
      "step": 8990
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13546110689640045,
      "learning_rate": 0.00038906329738278976,
      "loss": 1.754,
      "step": 9000
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.12389056384563446,
      "learning_rate": 0.0003890392046817023,
      "loss": 1.6444,
      "step": 9010
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13471394777297974,
      "learning_rate": 0.0003890150862202698,
      "loss": 1.6893,
      "step": 9020
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.14159949123859406,
      "learning_rate": 0.0003889909420017788,
      "loss": 1.685,
      "step": 9030
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13195070624351501,
      "learning_rate": 0.00038896677202951955,
      "loss": 1.6555,
      "step": 9040
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13802525401115417,
      "learning_rate": 0.00038894257630678564,
      "loss": 1.6851,
      "step": 9050
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.130794957280159,
      "learning_rate": 0.00038891835483687416,
      "loss": 1.6961,
      "step": 9060
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13027848303318024,
      "learning_rate": 0.0003888941076230858,
      "loss": 1.6602,
      "step": 9070
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.15656857192516327,
      "learning_rate": 0.0003888698346687249,
      "loss": 1.7005,
      "step": 9080
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13540861010551453,
      "learning_rate": 0.00038884553597709877,
      "loss": 1.6786,
      "step": 9090
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.12966085970401764,
      "learning_rate": 0.0003888212115515189,
      "loss": 1.7051,
      "step": 9100
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1450546681880951,
      "learning_rate": 0.0003887968613952999,
      "loss": 1.7378,
      "step": 9110
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13198581337928772,
      "learning_rate": 0.0003887724855117599,
      "loss": 1.7166,
      "step": 9120
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.14655737578868866,
      "learning_rate": 0.00038874808390422055,
      "loss": 1.7048,
      "step": 9130
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.14178557693958282,
      "learning_rate": 0.00038872365657600727,
      "loss": 1.7175,
      "step": 9140
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1497267335653305,
      "learning_rate": 0.0003886992035304485,
      "loss": 1.6875,
      "step": 9150
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.13873864710330963,
      "learning_rate": 0.00038867472477087666,
      "loss": 1.6518,
      "step": 9160
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.14022114872932434,
      "learning_rate": 0.00038865022030062745,
      "loss": 1.6726,
      "step": 9170
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.16763968765735626,
      "learning_rate": 0.0003886256901230399,
      "loss": 1.6726,
      "step": 9180
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.13509979844093323,
      "learning_rate": 0.00038860113424145696,
      "loss": 1.6489,
      "step": 9190
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.14755505323410034,
      "learning_rate": 0.00038857655265922477,
      "loss": 1.7358,
      "step": 9200
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1449791043996811,
      "learning_rate": 0.00038855194537969293,
      "loss": 1.729,
      "step": 9210
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.14036956429481506,
      "learning_rate": 0.0003885273124062149,
      "loss": 1.6459,
      "step": 9220
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.13039113581180573,
      "learning_rate": 0.00038850265374214726,
      "loss": 1.6881,
      "step": 9230
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.19317910075187683,
      "learning_rate": 0.00038847796939085033,
      "loss": 1.7277,
      "step": 9240
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1365085244178772,
      "learning_rate": 0.0003884532593556878,
      "loss": 1.6708,
      "step": 9250
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.14536632597446442,
      "learning_rate": 0.00038842852364002686,
      "loss": 1.6428,
      "step": 9260
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.17549118399620056,
      "learning_rate": 0.0003884037622472383,
      "loss": 1.7109,
      "step": 9270
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.16368693113327026,
      "learning_rate": 0.00038837897518069627,
      "loss": 1.659,
      "step": 9280
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1407514065504074,
      "learning_rate": 0.0003883541624437787,
      "loss": 1.6635,
      "step": 9290
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.18877118825912476,
      "learning_rate": 0.0003883293240398666,
      "loss": 1.7073,
      "step": 9300
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1376110315322876,
      "learning_rate": 0.0003883044599723448,
      "loss": 1.6782,
      "step": 9310
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1291997730731964,
      "learning_rate": 0.0003882795702446015,
      "loss": 1.6951,
      "step": 9320
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1362183392047882,
      "learning_rate": 0.0003882546548600284,
      "loss": 1.6561,
      "step": 9330
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1504468470811844,
      "learning_rate": 0.0003882297138220207,
      "loss": 1.7128,
      "step": 9340
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.15345141291618347,
      "learning_rate": 0.00038820474713397726,
      "loss": 1.7581,
      "step": 9350
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.15381696820259094,
      "learning_rate": 0.0003881797547993001,
      "loss": 1.6886,
      "step": 9360
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.14143483340740204,
      "learning_rate": 0.00038815473682139506,
      "loss": 1.6893,
      "step": 9370
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.13443134725093842,
      "learning_rate": 0.0003881296932036712,
      "loss": 1.7155,
      "step": 9380
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1517164260149002,
      "learning_rate": 0.00038810462394954135,
      "loss": 1.7083,
      "step": 9390
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.13738253712654114,
      "learning_rate": 0.0003880795290624216,
      "loss": 1.7058,
      "step": 9400
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1431797295808792,
      "learning_rate": 0.0003880544085457316,
      "loss": 1.6776,
      "step": 9410
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1337369680404663,
      "learning_rate": 0.00038802926240289463,
      "loss": 1.6781,
      "step": 9420
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1325959414243698,
      "learning_rate": 0.0003880040906373373,
      "loss": 1.7043,
      "step": 9430
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.13038308918476105,
      "learning_rate": 0.00038797889325248975,
      "loss": 1.6585,
      "step": 9440
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1321701556444168,
      "learning_rate": 0.0003879536702517856,
      "loss": 1.6258,
      "step": 9450
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1335831731557846,
      "learning_rate": 0.0003879284216386621,
      "loss": 1.6631,
      "step": 9460
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.13223014771938324,
      "learning_rate": 0.0003879031474165597,
      "loss": 1.6625,
      "step": 9470
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.152293398976326,
      "learning_rate": 0.00038787784758892267,
      "loss": 1.6885,
      "step": 9480
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.13212381303310394,
      "learning_rate": 0.0003878525221591985,
      "loss": 1.6607,
      "step": 9490
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.13295498490333557,
      "learning_rate": 0.0003878271711308384,
      "loss": 1.7179,
      "step": 9500
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.16751062870025635,
      "learning_rate": 0.00038780179450729677,
      "loss": 1.7262,
      "step": 9510
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.14195337891578674,
      "learning_rate": 0.00038777639229203186,
      "loss": 1.6958,
      "step": 9520
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1498609334230423,
      "learning_rate": 0.0003877509644885052,
      "loss": 1.7011,
      "step": 9530
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1556314378976822,
      "learning_rate": 0.00038772551110018177,
      "loss": 1.7051,
      "step": 9540
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.13929566740989685,
      "learning_rate": 0.00038770003213053005,
      "loss": 1.7249,
      "step": 9550
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1451740711927414,
      "learning_rate": 0.0003876745275830222,
      "loss": 1.7215,
      "step": 9560
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.14072318375110626,
      "learning_rate": 0.0003876489974611336,
      "loss": 1.6929,
      "step": 9570
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.15003930032253265,
      "learning_rate": 0.0003876234417683433,
      "loss": 1.6999,
      "step": 9580
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13815206289291382,
      "learning_rate": 0.0003875978605081338,
      "loss": 1.7009,
      "step": 9590
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.12934628129005432,
      "learning_rate": 0.00038757225368399095,
      "loss": 1.6663,
      "step": 9600
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13583825528621674,
      "learning_rate": 0.00038754662129940435,
      "loss": 1.6905,
      "step": 9610
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13525907695293427,
      "learning_rate": 0.0003875209633578667,
      "loss": 1.652,
      "step": 9620
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.138809934258461,
      "learning_rate": 0.00038749527986287454,
      "loss": 1.6523,
      "step": 9630
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.21345479786396027,
      "learning_rate": 0.0003874695708179278,
      "loss": 1.7747,
      "step": 9640
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1562160700559616,
      "learning_rate": 0.00038744383622652976,
      "loss": 1.6557,
      "step": 9650
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.16242921352386475,
      "learning_rate": 0.00038741807609218723,
      "loss": 1.7411,
      "step": 9660
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1379001885652542,
      "learning_rate": 0.0003873922904184107,
      "loss": 1.6967,
      "step": 9670
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.14510823786258698,
      "learning_rate": 0.0003873664792087138,
      "loss": 1.7457,
      "step": 9680
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13455097377300262,
      "learning_rate": 0.000387340642466614,
      "loss": 1.6778,
      "step": 9690
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.15125523507595062,
      "learning_rate": 0.0003873147801956319,
      "loss": 1.6823,
      "step": 9700
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1279924213886261,
      "learning_rate": 0.00038728889239929184,
      "loss": 1.6838,
      "step": 9710
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13603992760181427,
      "learning_rate": 0.00038726297908112155,
      "loss": 1.6892,
      "step": 9720
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.14827416837215424,
      "learning_rate": 0.0003872370402446522,
      "loss": 1.7074,
      "step": 9730
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13511140644550323,
      "learning_rate": 0.0003872110758934185,
      "loss": 1.6726,
      "step": 9740
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13389559090137482,
      "learning_rate": 0.0003871850860309585,
      "loss": 1.6368,
      "step": 9750
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1559256613254547,
      "learning_rate": 0.00038715907066081404,
      "loss": 1.6941,
      "step": 9760
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13027766346931458,
      "learning_rate": 0.0003871330297865301,
      "loss": 1.6764,
      "step": 9770
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13140283524990082,
      "learning_rate": 0.00038710696341165527,
      "loss": 1.6284,
      "step": 9780
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13621726632118225,
      "learning_rate": 0.00038708087153974157,
      "loss": 1.6927,
      "step": 9790
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.22002233564853668,
      "learning_rate": 0.0003870547541743446,
      "loss": 1.753,
      "step": 9800
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.163285031914711,
      "learning_rate": 0.00038702861131902343,
      "loss": 1.7628,
      "step": 9810
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.14639011025428772,
      "learning_rate": 0.00038700244297734047,
      "loss": 1.7623,
      "step": 9820
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1737416684627533,
      "learning_rate": 0.0003869762491528616,
      "loss": 1.7072,
      "step": 9830
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1327442079782486,
      "learning_rate": 0.00038695002984915646,
      "loss": 1.6497,
      "step": 9840
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.14944949746131897,
      "learning_rate": 0.0003869237850697977,
      "loss": 1.6907,
      "step": 9850
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1352148801088333,
      "learning_rate": 0.0003868975148183618,
      "loss": 1.7065,
      "step": 9860
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13450466096401215,
      "learning_rate": 0.0003868712190984287,
      "loss": 1.6629,
      "step": 9870
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.14856620132923126,
      "learning_rate": 0.0003868448979135816,
      "loss": 1.7225,
      "step": 9880
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1564270704984665,
      "learning_rate": 0.0003868185512674073,
      "loss": 1.7181,
      "step": 9890
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13846179842948914,
      "learning_rate": 0.00038679217916349606,
      "loss": 1.6527,
      "step": 9900
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13526107370853424,
      "learning_rate": 0.0003867657816054417,
      "loss": 1.65,
      "step": 9910
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.18866793811321259,
      "learning_rate": 0.0003867393585968412,
      "loss": 1.7315,
      "step": 9920
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1704215109348297,
      "learning_rate": 0.0003867129101412954,
      "loss": 1.6784,
      "step": 9930
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13856296241283417,
      "learning_rate": 0.00038668643624240837,
      "loss": 1.7344,
      "step": 9940
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1453276425600052,
      "learning_rate": 0.00038665993690378765,
      "loss": 1.6428,
      "step": 9950
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.14118853211402893,
      "learning_rate": 0.0003866334121290444,
      "loss": 1.6918,
      "step": 9960
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1492616832256317,
      "learning_rate": 0.00038660686192179306,
      "loss": 1.7261,
      "step": 9970
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1425030678510666,
      "learning_rate": 0.0003865802862856517,
      "loss": 1.7019,
      "step": 9980
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1411997377872467,
      "learning_rate": 0.0003865536852242417,
      "loss": 1.6625,
      "step": 9990
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1392057240009308,
      "learning_rate": 0.0003865270587411881,
      "loss": 1.6669,
      "step": 10000
    }
  ],
  "logging_steps": 10,
  "max_steps": 85104,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 1000,
  "total_flos": 8.458132783104e+18,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}