{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.25438041080051527,
  "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": 0.0003999999840334324,
      "loss": 8.1004,
      "step": 10
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.000399999936133732,
      "loss": 6.2987,
      "step": 20
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0003999998563009066,
      "loss": 6.0792,
      "step": 30
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00039999974453496884,
      "loss": 5.8398,
      "step": 40
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0003999996008359366,
      "loss": 5.5406,
      "step": 50
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0003999994252038328,
      "loss": 5.2281,
      "step": 60
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0003999992176386855,
      "loss": 5.0488,
      "step": 70
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00039999897814052787,
      "loss": 4.8797,
      "step": 80
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00039999870670939813,
      "loss": 4.726,
      "step": 90
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00039999840334533965,
      "loss": 4.5608,
      "step": 100
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0003999980680484007,
      "loss": 4.4583,
      "step": 110
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0003999977008186351,
      "loss": 4.4164,
      "step": 120
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0003999973016561012,
      "loss": 4.3347,
      "step": 130
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00039999687056086294,
      "loss": 4.2288,
      "step": 140
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00039999640753298903,
      "loss": 4.1609,
      "step": 150
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0003999959125725535,
      "loss": 4.0503,
      "step": 160
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00039999538567963525,
      "loss": 4.0077,
      "step": 170
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0003999948268543185,
      "loss": 3.9593,
      "step": 180
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00039999423609669247,
      "loss": 3.9097,
      "step": 190
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999936134068514,
      "loss": 3.8487,
      "step": 200
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999929587848948,
      "loss": 3.7907,
      "step": 210
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039999227223092726,
      "loss": 3.7483,
      "step": 220
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039999155374505826,
      "loss": 3.6961,
      "step": 230
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999908033274025,
      "loss": 3.5971,
      "step": 240
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999900209780799,
      "loss": 3.594,
      "step": 250
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999892066972154,
      "loss": 3.5695,
      "step": 260
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999883604849389,
      "loss": 3.4888,
      "step": 270
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999874823413855,
      "loss": 3.4582,
      "step": 280
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999865722666956,
      "loss": 3.4195,
      "step": 290
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999856302610142,
      "loss": 3.3806,
      "step": 300
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.000399984656324492,
      "loss": 3.3137,
      "step": 310
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999836504572844,
      "loss": 3.2644,
      "step": 320
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039998261265955195,
      "loss": 3.21,
      "step": 330
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039998154293146036,
      "loss": 3.1669,
      "step": 340
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999804412731805,
      "loss": 3.129,
      "step": 350
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039997930768488827,
      "loss": 3.0799,
      "step": 360
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999781421667645,
      "loss": 3.0365,
      "step": 370
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999769447189955,
      "loss": 3.0261,
      "step": 380
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999757153417723,
      "loss": 2.9473,
      "step": 390
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039997445403529134,
      "loss": 2.9481,
      "step": 400
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039997316079975386,
      "loss": 2.9903,
      "step": 410
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999718356353664,
      "loss": 2.9129,
      "step": 420
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999704785423406,
      "loss": 2.9139,
      "step": 430
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039996908952089305,
      "loss": 2.8713,
      "step": 440
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999676685712456,
      "loss": 2.8558,
      "step": 450
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039996621569362504,
      "loss": 2.8262,
      "step": 460
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999647308882635,
      "loss": 2.7535,
      "step": 470
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999632141553979,
      "loss": 2.7011,
      "step": 480
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039996166549527044,
      "loss": 2.7204,
      "step": 490
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999600849081285,
      "loss": 2.6671,
      "step": 500
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039995847239422417,
      "loss": 2.6731,
      "step": 510
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999568279538153,
      "loss": 2.673,
      "step": 520
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039995515158716417,
      "loss": 2.6212,
      "step": 530
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039995344329453854,
      "loss": 2.6267,
      "step": 540
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00039995170307621114,
      "loss": 2.5705,
      "step": 550
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999499309324598,
      "loss": 2.539,
      "step": 560
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999481268635675,
      "loss": 2.6173,
      "step": 570
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0003999462908698223,
      "loss": 2.5528,
      "step": 580
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039994442295151735,
      "loss": 2.5721,
      "step": 590
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039994252310895093,
      "loss": 2.5522,
      "step": 600
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003999405913424262,
      "loss": 2.5407,
      "step": 610
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003999386276522518,
      "loss": 2.4878,
      "step": 620
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039993663203874116,
      "loss": 2.4971,
      "step": 630
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039993460450221294,
      "loss": 2.5823,
      "step": 640
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039993254504299086,
      "loss": 2.4845,
      "step": 650
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039993045366140375,
      "loss": 2.4013,
      "step": 660
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003999283303577856,
      "loss": 2.4794,
      "step": 670
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039992617513247525,
      "loss": 2.3949,
      "step": 680
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039992398798581696,
      "loss": 2.4262,
      "step": 690
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039992176891815996,
      "loss": 2.3461,
      "step": 700
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039991951792985844,
      "loss": 2.3657,
      "step": 710
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039991723502127193,
      "loss": 2.3655,
      "step": 720
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039991492019276483,
      "loss": 2.407,
      "step": 730
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003999125734447068,
      "loss": 2.3978,
      "step": 740
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039991019477747254,
      "loss": 2.3706,
      "step": 750
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003999077841914418,
      "loss": 2.3245,
      "step": 760
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003999053416869995,
      "loss": 2.3302,
      "step": 770
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039990286726453557,
      "loss": 2.3197,
      "step": 780
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003999003609244452,
      "loss": 2.3013,
      "step": 790
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003998978226671284,
      "loss": 2.3447,
      "step": 800
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003998952524929906,
      "loss": 2.3617,
      "step": 810
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003998926504024421,
      "loss": 2.3264,
      "step": 820
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003998900163958984,
      "loss": 2.2853,
      "step": 830
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039988735047377996,
      "loss": 2.3353,
      "step": 840
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039988465263651263,
      "loss": 2.3734,
      "step": 850
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003998819228845269,
      "loss": 2.2988,
      "step": 860
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003998791612182589,
      "loss": 2.2447,
      "step": 870
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003998763676381493,
      "loss": 2.2734,
      "step": 880
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003998735421446444,
      "loss": 2.3543,
      "step": 890
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039987068473819516,
      "loss": 2.3313,
      "step": 900
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039986779541925784,
      "loss": 2.2926,
      "step": 910
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003998648741882938,
      "loss": 2.2219,
      "step": 920
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003998619210457695,
      "loss": 2.1825,
      "step": 930
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003998589359921563,
      "loss": 2.2279,
      "step": 940
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003998559190279309,
      "loss": 2.18,
      "step": 950
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003998528701535751,
      "loss": 2.2806,
      "step": 960
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039984978936957554,
      "loss": 2.2602,
      "step": 970
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00039984667667642424,
      "loss": 2.2162,
      "step": 980
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003998435320746181,
      "loss": 2.2006,
      "step": 990
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039984035556465925,
      "loss": 2.2247,
      "step": 1000
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039983714714705485,
      "loss": 2.2148,
      "step": 1010
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039983390682231726,
      "loss": 2.1869,
      "step": 1020
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039983063459096376,
      "loss": 2.1773,
      "step": 1030
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039982733045351677,
      "loss": 2.1144,
      "step": 1040
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039982399441050397,
      "loss": 2.1769,
      "step": 1050
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003998206264624579,
      "loss": 2.1617,
      "step": 1060
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039981722660991634,
      "loss": 2.1872,
      "step": 1070
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039981379485342223,
      "loss": 2.1406,
      "step": 1080
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039981033119352335,
      "loss": 2.1804,
      "step": 1090
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039980683563077286,
      "loss": 2.1605,
      "step": 1100
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003998033081657288,
      "loss": 2.159,
      "step": 1110
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003997997487989543,
      "loss": 2.1817,
      "step": 1120
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003997961575310179,
      "loss": 2.1418,
      "step": 1130
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039979253436249284,
      "loss": 2.166,
      "step": 1140
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003997888792939576,
      "loss": 2.1097,
      "step": 1150
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039978519232599584,
      "loss": 2.1546,
      "step": 1160
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039978147345919626,
      "loss": 2.1352,
      "step": 1170
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039977772269415255,
      "loss": 2.1723,
      "step": 1180
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039977394003146366,
      "loss": 2.1614,
      "step": 1190
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039977012547173346,
      "loss": 2.0975,
      "step": 1200
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039976627901557114,
      "loss": 2.1734,
      "step": 1210
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003997624006635907,
      "loss": 2.1047,
      "step": 1220
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039975849041641153,
      "loss": 2.1343,
      "step": 1230
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039975454827465777,
      "loss": 2.0768,
      "step": 1240
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039975057423895905,
      "loss": 2.1033,
      "step": 1250
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003997465683099497,
      "loss": 2.1109,
      "step": 1260
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039974253048826944,
      "loss": 2.1695,
      "step": 1270
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039973846077456305,
      "loss": 2.1097,
      "step": 1280
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039973435916948013,
      "loss": 2.0771,
      "step": 1290
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003997302256736757,
      "loss": 2.1272,
      "step": 1300
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039972606028780967,
      "loss": 2.1463,
      "step": 1310
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039972186301254713,
      "loss": 2.0335,
      "step": 1320
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039971763384855823,
      "loss": 2.0344,
      "step": 1330
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003997133727965183,
      "loss": 2.145,
      "step": 1340
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003997090798571076,
      "loss": 2.1313,
      "step": 1350
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039970475503101164,
      "loss": 2.1041,
      "step": 1360
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039970039831892083,
      "loss": 2.0418,
      "step": 1370
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003996960097215309,
      "loss": 2.0691,
      "step": 1380
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003996915892395425,
      "loss": 2.0488,
      "step": 1390
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039968713687366135,
      "loss": 2.104,
      "step": 1400
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003996826526245985,
      "loss": 2.107,
      "step": 1410
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039967813649306993,
      "loss": 2.023,
      "step": 1420
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039967358847979654,
      "loss": 2.1037,
      "step": 1430
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039966900858550466,
      "loss": 2.1022,
      "step": 1440
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003996643968109254,
      "loss": 2.1065,
      "step": 1450
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003996597531567953,
      "loss": 2.0959,
      "step": 1460
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039965507762385554,
      "loss": 2.0508,
      "step": 1470
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039965037021285287,
      "loss": 2.0902,
      "step": 1480
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039964563092453876,
      "loss": 2.0,
      "step": 1490
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003996408597596699,
      "loss": 2.0425,
      "step": 1500
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003996360567190082,
      "loss": 2.0393,
      "step": 1510
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003996312218033205,
      "loss": 2.0449,
      "step": 1520
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039962635501337876,
      "loss": 2.1379,
      "step": 1530
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039962145634995995,
      "loss": 2.1077,
      "step": 1540
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039961652581384634,
      "loss": 2.0143,
      "step": 1550
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039961156340582516,
      "loss": 2.0115,
      "step": 1560
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039960656912668864,
      "loss": 1.9883,
      "step": 1570
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003996015429772343,
      "loss": 2.0018,
      "step": 1580
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003995964849582646,
      "loss": 1.9963,
      "step": 1590
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039959139507058706,
      "loss": 1.9747,
      "step": 1600
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003995862733150145,
      "loss": 2.0209,
      "step": 1610
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039958111969236456,
      "loss": 2.05,
      "step": 1620
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039957593420346024,
      "loss": 2.0429,
      "step": 1630
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003995707168491293,
      "loss": 2.0028,
      "step": 1640
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003995654676302049,
      "loss": 1.9768,
      "step": 1650
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003995601865475252,
      "loss": 1.9517,
      "step": 1660
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003995548736019333,
      "loss": 1.9689,
      "step": 1670
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039954952879427754,
      "loss": 2.0619,
      "step": 1680
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003995441521254113,
      "loss": 2.0244,
      "step": 1690
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.000399538743596193,
      "loss": 2.0409,
      "step": 1700
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039953330320748623,
      "loss": 2.1296,
      "step": 1710
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039952783096015973,
      "loss": 2.0191,
      "step": 1720
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039952232685508713,
      "loss": 2.036,
      "step": 1730
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039951679089314724,
      "loss": 2.0275,
      "step": 1740
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.000399511223075224,
      "loss": 2.0407,
      "step": 1750
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003995056234022064,
      "loss": 1.9921,
      "step": 1760
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039949999187498844,
      "loss": 1.9966,
      "step": 1770
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003994943284944694,
      "loss": 1.9928,
      "step": 1780
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003994886332615535,
      "loss": 1.9647,
      "step": 1790
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003994829061771499,
      "loss": 1.9395,
      "step": 1800
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003994771472421733,
      "loss": 1.9352,
      "step": 1810
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039947135645754305,
      "loss": 1.9727,
      "step": 1820
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039946553382418375,
      "loss": 2.0184,
      "step": 1830
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039945967934302504,
      "loss": 1.9454,
      "step": 1840
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003994537930150018,
      "loss": 1.9974,
      "step": 1850
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039944787484105375,
      "loss": 1.9923,
      "step": 1860
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039944192482212584,
      "loss": 1.9374,
      "step": 1870
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003994359429591681,
      "loss": 2.035,
      "step": 1880
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003994299292531357,
      "loss": 1.9926,
      "step": 1890
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039942388370498873,
      "loss": 1.9723,
      "step": 1900
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003994178063156925,
      "loss": 1.9967,
      "step": 1910
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003994116970862173,
      "loss": 2.0542,
      "step": 1920
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003994055560175387,
      "loss": 2.0183,
      "step": 1930
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.000399399383110637,
      "loss": 2.0496,
      "step": 1940
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.000399393178366498,
      "loss": 1.9737,
      "step": 1950
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003993869417861123,
      "loss": 1.9379,
      "step": 1960
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003993806733704757,
      "loss": 2.0235,
      "step": 1970
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039937437312058903,
      "loss": 2.0181,
      "step": 1980
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039936804103745825,
      "loss": 2.0143,
      "step": 1990
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003993616771220944,
      "loss": 1.9753,
      "step": 2000
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003993552813755134,
      "loss": 1.9484,
      "step": 2010
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003993488537987366,
      "loss": 1.964,
      "step": 2020
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039934239439279025,
      "loss": 1.9522,
      "step": 2030
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039933590315870566,
      "loss": 2.0053,
      "step": 2040
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003993293800975193,
      "loss": 1.9453,
      "step": 2050
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039932282521027264,
      "loss": 1.8729,
      "step": 2060
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039931623849801223,
      "loss": 2.0346,
      "step": 2070
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003993096199617899,
      "loss": 1.9505,
      "step": 2080
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039930296960266224,
      "loss": 1.9834,
      "step": 2090
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003992962874216911,
      "loss": 1.9864,
      "step": 2100
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003992895734199434,
      "loss": 1.9417,
      "step": 2110
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039928282759849125,
      "loss": 1.9021,
      "step": 2120
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003992760499584116,
      "loss": 1.9185,
      "step": 2130
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003992692405007867,
      "loss": 1.9042,
      "step": 2140
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039926239922670367,
      "loss": 1.9236,
      "step": 2150
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00039925552613725496,
      "loss": 2.0219,
      "step": 2160
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039924862123353786,
      "loss": 1.9175,
      "step": 2170
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003992416845166549,
      "loss": 1.9339,
      "step": 2180
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003992347159877136,
      "loss": 1.8682,
      "step": 2190
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039922771564782663,
      "loss": 1.9337,
      "step": 2200
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003992206834981118,
      "loss": 1.9015,
      "step": 2210
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003992136195396916,
      "loss": 1.9687,
      "step": 2220
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039920652377369427,
      "loss": 1.9354,
      "step": 2230
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003991993962012525,
      "loss": 1.9228,
      "step": 2240
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039919223682350437,
      "loss": 1.9972,
      "step": 2250
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.000399185045641593,
      "loss": 1.9626,
      "step": 2260
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003991778226566667,
      "loss": 1.9724,
      "step": 2270
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039917056786987863,
      "loss": 1.8766,
      "step": 2280
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039916328128238704,
      "loss": 1.8419,
      "step": 2290
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003991559628953555,
      "loss": 1.855,
      "step": 2300
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003991486127099525,
      "loss": 1.966,
      "step": 2310
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003991412307273515,
      "loss": 1.9684,
      "step": 2320
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039913381694873113,
      "loss": 1.9364,
      "step": 2330
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003991263713752753,
      "loss": 1.9438,
      "step": 2340
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003991188940081727,
      "loss": 1.9081,
      "step": 2350
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039911138484861707,
      "loss": 1.8837,
      "step": 2360
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039910384389780764,
      "loss": 1.8935,
      "step": 2370
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003990962711569483,
      "loss": 1.8873,
      "step": 2380
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003990886666272481,
      "loss": 1.8894,
      "step": 2390
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003990810303099213,
      "loss": 1.9223,
      "step": 2400
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003990733622061872,
      "loss": 1.8759,
      "step": 2410
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003990656623172701,
      "loss": 1.9005,
      "step": 2420
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003990579306443993,
      "loss": 1.9132,
      "step": 2430
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039905016718880937,
      "loss": 1.8808,
      "step": 2440
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003990423719517399,
      "loss": 1.891,
      "step": 2450
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003990345449344355,
      "loss": 1.8811,
      "step": 2460
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003990266861381459,
      "loss": 1.8724,
      "step": 2470
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039901879556412576,
      "loss": 1.9121,
      "step": 2480
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039901087321363507,
      "loss": 1.9112,
      "step": 2490
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003990029190879387,
      "loss": 1.931,
      "step": 2500
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039899493318830664,
      "loss": 1.8886,
      "step": 2510
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039898691551601396,
      "loss": 1.953,
      "step": 2520
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003989788660723408,
      "loss": 1.9476,
      "step": 2530
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00039897078485857257,
      "loss": 1.916,
      "step": 2540
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003989626718759993,
      "loss": 1.9513,
      "step": 2550
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039895452712591646,
      "loss": 1.9378,
      "step": 2560
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003989463506096244,
      "loss": 1.9231,
      "step": 2570
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039893814232842877,
      "loss": 1.8928,
      "step": 2580
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003989299022836401,
      "loss": 1.8642,
      "step": 2590
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039892163047657405,
      "loss": 1.8858,
      "step": 2600
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003989133269085513,
      "loss": 1.9193,
      "step": 2610
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003989049915808977,
      "loss": 1.8847,
      "step": 2620
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003988966244949441,
      "loss": 1.9073,
      "step": 2630
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003988882256520264,
      "loss": 1.9436,
      "step": 2640
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039887979505348565,
      "loss": 1.9591,
      "step": 2650
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003988713327006679,
      "loss": 1.8816,
      "step": 2660
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003988628385949242,
      "loss": 1.9548,
      "step": 2670
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039885431273761095,
      "loss": 1.9138,
      "step": 2680
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003988457551300894,
      "loss": 2.158,
      "step": 2690
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039883716577372583,
      "loss": 2.0845,
      "step": 2700
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039882854466989176,
      "loss": 2.0201,
      "step": 2710
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039881989181996354,
      "loss": 1.9172,
      "step": 2720
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039881120722532284,
      "loss": 1.9177,
      "step": 2730
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003988024908873563,
      "loss": 1.9243,
      "step": 2740
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003987937428074556,
      "loss": 1.9129,
      "step": 2750
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039878496298701745,
      "loss": 1.8875,
      "step": 2760
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039877615142744376,
      "loss": 1.8674,
      "step": 2770
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003987673081301414,
      "loss": 1.913,
      "step": 2780
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039875843309652233,
      "loss": 1.9136,
      "step": 2790
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003987495263280037,
      "loss": 1.8634,
      "step": 2800
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039874058782600745,
      "loss": 1.8743,
      "step": 2810
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039873161759196085,
      "loss": 1.7969,
      "step": 2820
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039872261562729606,
      "loss": 1.9213,
      "step": 2830
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039871358193345046,
      "loss": 1.8824,
      "step": 2840
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003987045165118664,
      "loss": 1.9052,
      "step": 2850
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039869541936399136,
      "loss": 1.8925,
      "step": 2860
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003986862904912777,
      "loss": 1.8846,
      "step": 2870
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003986771298951832,
      "loss": 1.8653,
      "step": 2880
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003986679375771703,
      "loss": 1.8782,
      "step": 2890
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00039865871353870683,
      "loss": 1.959,
      "step": 2900
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003986494577812655,
      "loss": 1.8541,
      "step": 2910
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003986401703063241,
      "loss": 1.8953,
      "step": 2920
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003986308511153656,
      "loss": 1.9349,
      "step": 2930
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003986215002098778,
      "loss": 1.8454,
      "step": 2940
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039861211759135393,
      "loss": 1.8725,
      "step": 2950
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.000398602703261292,
      "loss": 1.9361,
      "step": 2960
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.000398593257221195,
      "loss": 1.886,
      "step": 2970
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039858377947257133,
      "loss": 1.9202,
      "step": 2980
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003985742700169342,
      "loss": 1.9123,
      "step": 2990
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039856472885580195,
      "loss": 1.9366,
      "step": 3000
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.000398555155990698,
      "loss": 1.8829,
      "step": 3010
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039854555142315067,
      "loss": 1.9275,
      "step": 3020
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039853591515469366,
      "loss": 1.9413,
      "step": 3030
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039852624718686544,
      "loss": 1.8638,
      "step": 3040
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039851654752120976,
      "loss": 1.9021,
      "step": 3050
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039850681615927517,
      "loss": 1.9018,
      "step": 3060
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039849705310261556,
      "loss": 1.8719,
      "step": 3070
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039848725835278973,
      "loss": 1.8713,
      "step": 3080
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039847743191136144,
      "loss": 1.8903,
      "step": 3090
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039846757377989985,
      "loss": 1.8186,
      "step": 3100
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039845768395997883,
      "loss": 1.8552,
      "step": 3110
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039844776245317743,
      "loss": 1.8805,
      "step": 3120
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003984378092610799,
      "loss": 1.9612,
      "step": 3130
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039842782438527524,
      "loss": 1.8515,
      "step": 3140
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003984178078273579,
      "loss": 1.8468,
      "step": 3150
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039840775958892695,
      "loss": 1.8468,
      "step": 3160
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039839767967158696,
      "loss": 1.8479,
      "step": 3170
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039838756807694724,
      "loss": 1.8419,
      "step": 3180
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003983774248066222,
      "loss": 1.8238,
      "step": 3190
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003983672498622316,
      "loss": 1.8248,
      "step": 3200
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003983570432453998,
      "loss": 1.8572,
      "step": 3210
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039834680495775654,
      "loss": 1.9093,
      "step": 3220
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039833653500093646,
      "loss": 1.898,
      "step": 3230
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039832623337657946,
      "loss": 1.847,
      "step": 3240
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039831590008633023,
      "loss": 1.8168,
      "step": 3250
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003983055351318387,
      "loss": 1.8661,
      "step": 3260
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039829513851475973,
      "loss": 1.882,
      "step": 3270
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039828471023675336,
      "loss": 1.8127,
      "step": 3280
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003982742502994847,
      "loss": 1.8136,
      "step": 3290
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039826375870462364,
      "loss": 1.8412,
      "step": 3300
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039825323545384555,
      "loss": 1.8997,
      "step": 3310
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039824268054883044,
      "loss": 1.8213,
      "step": 3320
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003982320939912637,
      "loss": 1.8443,
      "step": 3330
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00039822147578283564,
      "loss": 1.8576,
      "step": 3340
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003982108259252415,
      "loss": 1.8235,
      "step": 3350
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003982001444201818,
      "loss": 1.8112,
      "step": 3360
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039818943126936204,
      "loss": 1.8501,
      "step": 3370
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003981786864744926,
      "loss": 1.9042,
      "step": 3380
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039816791003728923,
      "loss": 1.8704,
      "step": 3390
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039815710195947244,
      "loss": 1.8226,
      "step": 3400
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039814626224276797,
      "loss": 1.8682,
      "step": 3410
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039813539088890646,
      "loss": 1.822,
      "step": 3420
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003981244878996238,
      "loss": 1.8546,
      "step": 3430
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003981135532766607,
      "loss": 1.8375,
      "step": 3440
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039810258702176323,
      "loss": 1.9318,
      "step": 3450
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039809158913668214,
      "loss": 1.8846,
      "step": 3460
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003980805596231735,
      "loss": 1.841,
      "step": 3470
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003980694984829984,
      "loss": 1.7964,
      "step": 3480
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039805840571792276,
      "loss": 1.8562,
      "step": 3490
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003980472813297179,
      "loss": 1.8546,
      "step": 3500
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039803612532015984,
      "loss": 1.8304,
      "step": 3510
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039802493769102994,
      "loss": 1.8223,
      "step": 3520
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003980137184441143,
      "loss": 1.8645,
      "step": 3530
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003980024675812046,
      "loss": 1.8605,
      "step": 3540
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003979911851040968,
      "loss": 1.8072,
      "step": 3550
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003979798710145926,
      "loss": 1.8886,
      "step": 3560
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003979685253144984,
      "loss": 1.7982,
      "step": 3570
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003979571480056257,
      "loss": 1.8072,
      "step": 3580
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003979457390897911,
      "loss": 1.8435,
      "step": 3590
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003979342985688161,
      "loss": 1.832,
      "step": 3600
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003979228264445275,
      "loss": 1.8205,
      "step": 3610
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003979113227187569,
      "loss": 1.8369,
      "step": 3620
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003978997873933412,
      "loss": 1.8123,
      "step": 3630
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.000397888220470122,
      "loss": 1.7854,
      "step": 3640
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003978766219509463,
      "loss": 1.7781,
      "step": 3650
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003978649918376658,
      "loss": 1.8043,
      "step": 3660
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039785333013213765,
      "loss": 1.8317,
      "step": 3670
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039784163683622374,
      "loss": 1.7834,
      "step": 3680
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003978299119517911,
      "loss": 1.8493,
      "step": 3690
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00039781815548071167,
      "loss": 1.7785,
      "step": 3700
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003978063674248627,
      "loss": 1.8433,
      "step": 3710
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003977945477861263,
      "loss": 1.7909,
      "step": 3720
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003977826965663896,
      "loss": 1.83,
      "step": 3730
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039777081376754487,
      "loss": 1.7919,
      "step": 3740
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039775889939148946,
      "loss": 1.7883,
      "step": 3750
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003977469534401256,
      "loss": 1.7908,
      "step": 3760
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039773497591536063,
      "loss": 1.7429,
      "step": 3770
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.000397722966819107,
      "loss": 1.8088,
      "step": 3780
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039771092615328217,
      "loss": 1.8726,
      "step": 3790
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003976988539198086,
      "loss": 1.8244,
      "step": 3800
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003976867501206138,
      "loss": 1.8047,
      "step": 3810
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003976746147576303,
      "loss": 1.8664,
      "step": 3820
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003976624478327958,
      "loss": 1.8955,
      "step": 3830
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039765024934805283,
      "loss": 1.8355,
      "step": 3840
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039763801930534917,
      "loss": 1.8964,
      "step": 3850
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039762575770663737,
      "loss": 1.9149,
      "step": 3860
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039761346455387535,
      "loss": 1.878,
      "step": 3870
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003976011398490259,
      "loss": 1.7572,
      "step": 3880
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039758878359405676,
      "loss": 1.786,
      "step": 3890
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039757639579094084,
      "loss": 1.8305,
      "step": 3900
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003975639764416561,
      "loss": 1.7545,
      "step": 3910
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039755152554818543,
      "loss": 1.7633,
      "step": 3920
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003975390431125168,
      "loss": 1.7908,
      "step": 3930
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003975265291366432,
      "loss": 1.8037,
      "step": 3940
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003975139836225628,
      "loss": 1.8299,
      "step": 3950
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039750140657227853,
      "loss": 1.7984,
      "step": 3960
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039748879798779866,
      "loss": 1.8757,
      "step": 3970
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003974761578711363,
      "loss": 1.764,
      "step": 3980
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003974634862243096,
      "loss": 1.7753,
      "step": 3990
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003974507830493418,
      "loss": 1.7783,
      "step": 4000
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003974380483482612,
      "loss": 1.8552,
      "step": 4010
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003974252821231011,
      "loss": 1.8819,
      "step": 4020
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003974124843758998,
      "loss": 1.8257,
      "step": 4030
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003973996551087006,
      "loss": 1.8212,
      "step": 4040
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039738679432355193,
      "loss": 1.84,
      "step": 4050
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003973739020225073,
      "loss": 1.8493,
      "step": 4060
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003973609782076251,
      "loss": 1.8039,
      "step": 4070
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003973480228809689,
      "loss": 1.7981,
      "step": 4080
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039733503604460704,
      "loss": 1.8088,
      "step": 4090
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003973220177006132,
      "loss": 1.7446,
      "step": 4100
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039730896785106597,
      "loss": 1.8402,
      "step": 4110
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00039729588649804894,
      "loss": 1.799,
      "step": 4120
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003972827736436507,
      "loss": 1.8064,
      "step": 4130
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039726962928996503,
      "loss": 1.7946,
      "step": 4140
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039725645343909055,
      "loss": 1.8193,
      "step": 4150
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039724324609313104,
      "loss": 1.7728,
      "step": 4160
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039723000725419517,
      "loss": 1.8646,
      "step": 4170
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039721673692439687,
      "loss": 1.8326,
      "step": 4180
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039720343510585483,
      "loss": 1.8098,
      "step": 4190
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039719010180069294,
      "loss": 1.8529,
      "step": 4200
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003971767370110401,
      "loss": 1.811,
      "step": 4210
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039716334073903016,
      "loss": 1.8112,
      "step": 4220
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003971499129868021,
      "loss": 1.7949,
      "step": 4230
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039713645375649985,
      "loss": 1.8224,
      "step": 4240
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003971229630502723,
      "loss": 1.8517,
      "step": 4250
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003971094408702736,
      "loss": 1.8168,
      "step": 4260
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039709588721866267,
      "loss": 1.7842,
      "step": 4270
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039708230209760365,
      "loss": 1.8217,
      "step": 4280
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003970686855092655,
      "loss": 1.7968,
      "step": 4290
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003970550374558224,
      "loss": 1.7345,
      "step": 4300
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003970413579394535,
      "loss": 1.8,
      "step": 4310
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003970276469623429,
      "loss": 1.808,
      "step": 4320
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003970139045266798,
      "loss": 1.8049,
      "step": 4330
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003970001306346583,
      "loss": 1.8013,
      "step": 4340
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003969863252884778,
      "loss": 1.8282,
      "step": 4350
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003969724884903424,
      "loss": 1.8122,
      "step": 4360
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003969586202424614,
      "loss": 1.8082,
      "step": 4370
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003969447205470491,
      "loss": 1.7345,
      "step": 4380
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039693078940632474,
      "loss": 1.7867,
      "step": 4390
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003969168268225127,
      "loss": 1.7596,
      "step": 4400
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003969028327978424,
      "loss": 1.8147,
      "step": 4410
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003968888073345481,
      "loss": 1.8479,
      "step": 4420
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039687475043486916,
      "loss": 1.7887,
      "step": 4430
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039686066210105006,
      "loss": 1.8487,
      "step": 4440
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039684654233534017,
      "loss": 1.8783,
      "step": 4450
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039683239113999394,
      "loss": 1.7634,
      "step": 4460
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003968182085172709,
      "loss": 1.7534,
      "step": 4470
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003968039944694355,
      "loss": 1.7874,
      "step": 4480
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039678974899875715,
      "loss": 1.7343,
      "step": 4490
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003967754721075105,
      "loss": 1.7943,
      "step": 4500
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00039676116379797494,
      "loss": 1.7694,
      "step": 4510
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0003967468240724351,
      "loss": 1.823,
      "step": 4520
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003967324529331805,
      "loss": 1.8147,
      "step": 4530
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003967180503825058,
      "loss": 1.779,
      "step": 4540
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003967036164227105,
      "loss": 1.7434,
      "step": 4550
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039668915105609925,
      "loss": 1.8157,
      "step": 4560
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003966746542849816,
      "loss": 1.7972,
      "step": 4570
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003966601261116723,
      "loss": 1.768,
      "step": 4580
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.000396645566538491,
      "loss": 1.7521,
      "step": 4590
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003966309755677623,
      "loss": 1.8067,
      "step": 4600
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003966163532018158,
      "loss": 1.7907,
      "step": 4610
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003966016994429864,
      "loss": 1.7259,
      "step": 4620
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003965870142936136,
      "loss": 1.8267,
      "step": 4630
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003965722977560422,
      "loss": 1.7433,
      "step": 4640
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.000396557549832622,
      "loss": 1.7433,
      "step": 4650
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003965427705257076,
      "loss": 1.8058,
      "step": 4660
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003965279598376588,
      "loss": 1.8024,
      "step": 4670
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003965131177708404,
      "loss": 1.7884,
      "step": 4680
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003964982443276221,
      "loss": 1.7835,
      "step": 4690
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003964833395103788,
      "loss": 1.8312,
      "step": 4700
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003964684033214901,
      "loss": 1.7961,
      "step": 4710
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.000396453435763341,
      "loss": 1.7441,
      "step": 4720
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039643843683832115,
      "loss": 1.7585,
      "step": 4730
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039642340654882544,
      "loss": 1.7572,
      "step": 4740
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039640834489725366,
      "loss": 1.8212,
      "step": 4750
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003963932518860106,
      "loss": 1.741,
      "step": 4760
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039637812751750623,
      "loss": 1.7815,
      "step": 4770
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003963629717941553,
      "loss": 1.7903,
      "step": 4780
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039634778471837764,
      "loss": 1.768,
      "step": 4790
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003963325662925981,
      "loss": 1.7375,
      "step": 4800
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039631731651924666,
      "loss": 1.7954,
      "step": 4810
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.000396302035400758,
      "loss": 1.7816,
      "step": 4820
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003962867229395721,
      "loss": 1.8233,
      "step": 4830
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003962713791381338,
      "loss": 1.7717,
      "step": 4840
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003962560039988931,
      "loss": 1.8,
      "step": 4850
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039624059752430473,
      "loss": 1.7848,
      "step": 4860
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003962251597168286,
      "loss": 1.7773,
      "step": 4870
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00039620969057892967,
      "loss": 1.8012,
      "step": 4880
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003961941901130778,
      "loss": 1.7854,
      "step": 4890
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003961786583217478,
      "loss": 1.7833,
      "step": 4900
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0003961630952074197,
      "loss": 1.7826,
      "step": 4910
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039614750077257826,
      "loss": 1.8489,
      "step": 4920
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039613187501971346,
      "loss": 1.8329,
      "step": 4930
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003961162179513202,
      "loss": 1.7486,
      "step": 4940
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003961005295698984,
      "loss": 1.7631,
      "step": 4950
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039608480987795284,
      "loss": 1.7728,
      "step": 4960
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003960690588779935,
      "loss": 1.7824,
      "step": 4970
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003960532765725352,
      "loss": 1.7664,
      "step": 4980
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.000396037462964098,
      "loss": 1.7976,
      "step": 4990
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039602161805520666,
      "loss": 1.7426,
      "step": 5000
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003960057418483911,
      "loss": 1.806,
      "step": 5010
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003959898343461862,
      "loss": 1.7553,
      "step": 5020
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003959738955511318,
      "loss": 1.7616,
      "step": 5030
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039595792546577276,
      "loss": 1.767,
      "step": 5040
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039594192409265913,
      "loss": 1.762,
      "step": 5050
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039592589143434565,
      "loss": 1.8521,
      "step": 5060
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003959098274933921,
      "loss": 2.1704,
      "step": 5070
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039589373227236354,
      "loss": 1.8074,
      "step": 5080
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003958776057738297,
      "loss": 1.854,
      "step": 5090
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039586144800036544,
      "loss": 1.7474,
      "step": 5100
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003958452589545506,
      "loss": 1.7693,
      "step": 5110
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039582903863897,
      "loss": 1.7798,
      "step": 5120
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039581278705621355,
      "loss": 1.7063,
      "step": 5130
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.000395796504208876,
      "loss": 1.7747,
      "step": 5140
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039578019009955717,
      "loss": 1.7526,
      "step": 5150
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003957638447308619,
      "loss": 1.7272,
      "step": 5160
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003957474681053999,
      "loss": 1.727,
      "step": 5170
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.000395731060225786,
      "loss": 1.7858,
      "step": 5180
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039571462109464005,
      "loss": 1.7879,
      "step": 5190
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003956981507145867,
      "loss": 1.7494,
      "step": 5200
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003956816490882558,
      "loss": 1.7716,
      "step": 5210
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039566511621828203,
      "loss": 1.7585,
      "step": 5220
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039564855210730515,
      "loss": 1.7309,
      "step": 5230
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003956319567579698,
      "loss": 1.7503,
      "step": 5240
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003956153301729258,
      "loss": 1.7631,
      "step": 5250
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039559867235482784,
      "loss": 1.6883,
      "step": 5260
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039558198330633555,
      "loss": 1.7946,
      "step": 5270
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039556526303011354,
      "loss": 1.7962,
      "step": 5280
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00039554851152883157,
      "loss": 1.7869,
      "step": 5290
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0003955317288051643,
      "loss": 1.7507,
      "step": 5300
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003955149148617912,
      "loss": 1.7614,
      "step": 5310
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003954980697013971,
      "loss": 1.7288,
      "step": 5320
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003954811933266714,
      "loss": 1.8662,
      "step": 5330
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003954642857403088,
      "loss": 1.7724,
      "step": 5340
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039544734694500874,
      "loss": 1.8028,
      "step": 5350
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003954303769434759,
      "loss": 1.7507,
      "step": 5360
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003954133757384197,
      "loss": 1.74,
      "step": 5370
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003953963433325547,
      "loss": 1.7622,
      "step": 5380
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039537927972860043,
      "loss": 1.7755,
      "step": 5390
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003953621849292813,
      "loss": 1.7695,
      "step": 5400
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039534505893732677,
      "loss": 1.7455,
      "step": 5410
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003953279017554713,
      "loss": 1.8059,
      "step": 5420
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003953107133864542,
      "loss": 1.7896,
      "step": 5430
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039529349383302006,
      "loss": 1.7552,
      "step": 5440
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039527624309791806,
      "loss": 1.7444,
      "step": 5450
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039525896118390266,
      "loss": 1.714,
      "step": 5460
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039524164809373315,
      "loss": 1.8,
      "step": 5470
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003952243038301738,
      "loss": 1.7523,
      "step": 5480
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.000395206928395994,
      "loss": 1.7239,
      "step": 5490
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039518952179396795,
      "loss": 1.7658,
      "step": 5500
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039517208402687487,
      "loss": 1.7425,
      "step": 5510
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039515461509749897,
      "loss": 1.7497,
      "step": 5520
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039513711500862946,
      "loss": 1.7329,
      "step": 5530
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039511958376306055,
      "loss": 1.7289,
      "step": 5540
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003951020213635912,
      "loss": 1.7602,
      "step": 5550
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039508442781302576,
      "loss": 1.8221,
      "step": 5560
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003950668031141732,
      "loss": 1.8125,
      "step": 5570
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039504914726984754,
      "loss": 1.7633,
      "step": 5580
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039503146028286787,
      "loss": 1.7667,
      "step": 5590
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003950137421560582,
      "loss": 1.7723,
      "step": 5600
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003949959928922475,
      "loss": 1.7784,
      "step": 5610
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003949782124942697,
      "loss": 1.7889,
      "step": 5620
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039496040096496373,
      "loss": 1.776,
      "step": 5630
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039494255830717346,
      "loss": 1.7307,
      "step": 5640
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039492468452374775,
      "loss": 1.7699,
      "step": 5650
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039490677961754044,
      "loss": 1.7956,
      "step": 5660
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003948888435914104,
      "loss": 1.7993,
      "step": 5670
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00039487087644822126,
      "loss": 1.7359,
      "step": 5680
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003948528781908419,
      "loss": 1.7289,
      "step": 5690
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0003948348488221459,
      "loss": 1.8222,
      "step": 5700
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039481678834501203,
      "loss": 1.768,
      "step": 5710
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039479869676232386,
      "loss": 1.7411,
      "step": 5720
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039478057407697,
      "loss": 1.7199,
      "step": 5730
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.000394762420291844,
      "loss": 1.7037,
      "step": 5740
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003947442354098445,
      "loss": 1.7107,
      "step": 5750
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039472601943387495,
      "loss": 1.7102,
      "step": 5760
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039470777236684377,
      "loss": 1.7865,
      "step": 5770
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039468949421166436,
      "loss": 1.7583,
      "step": 5780
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003946711849712553,
      "loss": 1.7925,
      "step": 5790
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003946528446485398,
      "loss": 1.7322,
      "step": 5800
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039463447324644614,
      "loss": 1.7248,
      "step": 5810
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039461607076790773,
      "loss": 1.7555,
      "step": 5820
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039459763721586265,
      "loss": 1.7371,
      "step": 5830
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003945791725932543,
      "loss": 1.713,
      "step": 5840
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003945606769030307,
      "loss": 1.7754,
      "step": 5850
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039454215014814506,
      "loss": 1.7139,
      "step": 5860
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003945235923315554,
      "loss": 1.7106,
      "step": 5870
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039450500345622485,
      "loss": 1.7615,
      "step": 5880
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003944863835251214,
      "loss": 1.7612,
      "step": 5890
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003944677325412179,
      "loss": 1.7308,
      "step": 5900
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003944490505074924,
      "loss": 1.7304,
      "step": 5910
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039443033742692774,
      "loss": 1.7037,
      "step": 5920
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039441159330251167,
      "loss": 1.725,
      "step": 5930
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003943928181372372,
      "loss": 1.7284,
      "step": 5940
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039437401193410183,
      "loss": 1.7352,
      "step": 5950
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003943551746961084,
      "loss": 1.7093,
      "step": 5960
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003943363064262646,
      "loss": 1.8308,
      "step": 5970
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003943174071275829,
      "loss": 2.0842,
      "step": 5980
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.000394298476803081,
      "loss": 1.7864,
      "step": 5990
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003942795154557814,
      "loss": 1.6993,
      "step": 6000
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003942605230887116,
      "loss": 1.7407,
      "step": 6010
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039424149970490396,
      "loss": 1.695,
      "step": 6020
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039422244530739584,
      "loss": 1.7297,
      "step": 6030
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003942033598992297,
      "loss": 1.7075,
      "step": 6040
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003941842434834527,
      "loss": 1.7377,
      "step": 6050
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003941650960631172,
      "loss": 1.7527,
      "step": 6060
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003941459176412802,
      "loss": 1.7288,
      "step": 6070
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00039412670822100405,
      "loss": 1.7208,
      "step": 6080
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0003941074678053557,
      "loss": 1.7281,
      "step": 6090
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003940881963974072,
      "loss": 1.7372,
      "step": 6100
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039406889400023557,
      "loss": 1.767,
      "step": 6110
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039404956061692267,
      "loss": 1.6922,
      "step": 6120
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003940301962505555,
      "loss": 1.6791,
      "step": 6130
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039401080090422573,
      "loss": 1.7537,
      "step": 6140
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039399137458103026,
      "loss": 1.7717,
      "step": 6150
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039397191728407076,
      "loss": 1.7228,
      "step": 6160
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003939524290164539,
      "loss": 1.7459,
      "step": 6170
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039393290978129126,
      "loss": 1.7059,
      "step": 6180
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003939133595816994,
      "loss": 1.7093,
      "step": 6190
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039389377842079986,
      "loss": 1.737,
      "step": 6200
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039387416630171904,
      "loss": 1.7162,
      "step": 6210
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039385452322758833,
      "loss": 1.7225,
      "step": 6220
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039383484920154407,
      "loss": 1.73,
      "step": 6230
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039381514422672745,
      "loss": 1.7722,
      "step": 6240
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003937954083062848,
      "loss": 1.7125,
      "step": 6250
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039377564144336713,
      "loss": 1.7239,
      "step": 6260
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039375584364113067,
      "loss": 1.7147,
      "step": 6270
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003937360149027364,
      "loss": 1.7597,
      "step": 6280
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039371615523135024,
      "loss": 1.7395,
      "step": 6290
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003936962646301432,
      "loss": 1.6462,
      "step": 6300
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003936763431022909,
      "loss": 1.738,
      "step": 6310
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039365639065097445,
      "loss": 1.7236,
      "step": 6320
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039363640727937927,
      "loss": 1.6844,
      "step": 6330
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003936163929906963,
      "loss": 1.7056,
      "step": 6340
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039359634778812086,
      "loss": 1.7279,
      "step": 6350
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039357627167485365,
      "loss": 1.7891,
      "step": 6360
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003935561646541001,
      "loss": 1.8111,
      "step": 6370
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039353602672907067,
      "loss": 1.7169,
      "step": 6380
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003935158579029806,
      "loss": 1.7333,
      "step": 6390
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003934956581790501,
      "loss": 1.7419,
      "step": 6400
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039347542756050453,
      "loss": 1.7139,
      "step": 6410
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039345516605057397,
      "loss": 1.7196,
      "step": 6420
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039343487365249346,
      "loss": 1.6798,
      "step": 6430
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003934145503695031,
      "loss": 1.7796,
      "step": 6440
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003933941962048476,
      "loss": 1.7407,
      "step": 6450
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039337381116177705,
      "loss": 1.68,
      "step": 6460
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0003933533952435461,
      "loss": 1.7518,
      "step": 6470
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00039333294845341453,
      "loss": 1.7048,
      "step": 6480
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.000393312470794647,
      "loss": 1.746,
      "step": 6490
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0003932919622705131,
      "loss": 1.7358,
      "step": 6500
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00039327142288428726,
      "loss": 1.7378,
      "step": 6510
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.000393250852639249,
      "loss": 1.7156,
      "step": 6520
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0003932302515386826,
      "loss": 1.7146,
      "step": 6530
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00039320961958587745,
      "loss": 1.7649,
      "step": 6540
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00039318895678412766,
      "loss": 1.7608,
      "step": 6550
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0003931682631367324,
      "loss": 1.6887,
      "step": 6560
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0003931475386469958,
      "loss": 1.7031,
      "step": 6570
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00039312678331822684,
      "loss": 1.7654,
      "step": 6580
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0003931059971537394,
      "loss": 1.7527,
      "step": 6590
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00039308518015685227,
      "loss": 1.7217,
      "step": 6600
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00039306433233088925,
      "loss": 1.7602,
      "step": 6610
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.000393043453679179,
      "loss": 1.7707,
      "step": 6620
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0003930225442050552,
      "loss": 1.7279,
      "step": 6630
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00039300160391185637,
      "loss": 1.7004,
      "step": 6640
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0003929806328029258,
      "loss": 1.7322,
      "step": 6650
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00039295963088161205,
      "loss": 1.7124,
      "step": 6660
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00039293859815126833,
      "loss": 1.7329,
      "step": 6670
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0003929175346152528,
      "loss": 1.7658,
      "step": 6680
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00039289644027692863,
      "loss": 1.6922,
      "step": 6690
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00039287531513966387,
      "loss": 1.7263,
      "step": 6700
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00039285415920683146,
      "loss": 1.7444,
      "step": 6710
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00039283297248180924,
      "loss": 1.7123,
      "step": 6720
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00039281175496798013,
      "loss": 1.7342,
      "step": 6730
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0003927905066687317,
      "loss": 1.7511,
      "step": 6740
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00039276922758745665,
      "loss": 1.6741,
      "step": 6750
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0003927479177275525,
      "loss": 1.6909,
      "step": 6760
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0003927265770924216,
      "loss": 1.7151,
      "step": 6770
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00039270520568547156,
      "loss": 1.7389,
      "step": 6780
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00039268380351011446,
      "loss": 1.7088,
      "step": 6790
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00039266237056976753,
      "loss": 1.7261,
      "step": 6800
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00039264090686785293,
      "loss": 1.7018,
      "step": 6810
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00039261941240779756,
      "loss": 1.6874,
      "step": 6820
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0003925978871930335,
      "loss": 1.7276,
      "step": 6830
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00039257633122699755,
      "loss": 1.7157,
      "step": 6840
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0003925547445131313,
      "loss": 1.7039,
      "step": 6850
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0003925331270548816,
      "loss": 1.7528,
      "step": 6860
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0003925114788557,
      "loss": 1.6982,
      "step": 6870
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0003924897999190429,
      "loss": 1.7167,
      "step": 6880
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00039246809024837164,
      "loss": 1.6949,
      "step": 6890
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00039244634984715257,
      "loss": 1.8046,
      "step": 6900
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00039242457871885696,
      "loss": 1.7028,
      "step": 6910
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0003924027768669608,
      "loss": 1.7132,
      "step": 6920
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0003923809442949452,
      "loss": 1.6916,
      "step": 6930
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0003923590810062959,
      "loss": 1.7176,
      "step": 6940
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00039233718700450393,
      "loss": 1.7359,
      "step": 6950
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00039231526229306483,
      "loss": 1.7271,
      "step": 6960
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00039229330687547934,
      "loss": 1.6948,
      "step": 6970
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00039227132075525295,
      "loss": 1.7418,
      "step": 6980
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0003922493039358961,
      "loss": 1.7207,
      "step": 6990
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0003922272564209241,
      "loss": 1.6874,
      "step": 7000
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00039220517821385715,
      "loss": 1.7025,
      "step": 7010
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00039218306931822043,
      "loss": 1.7748,
      "step": 7020
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.000392160929737544,
      "loss": 1.7008,
      "step": 7030
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0003921387594753627,
      "loss": 1.6817,
      "step": 7040
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1620354801416397,
      "learning_rate": 0.0003921165585352165,
      "loss": 1.7891,
      "step": 7050
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.148910254240036,
      "learning_rate": 0.00039209432692064995,
      "loss": 1.7461,
      "step": 7060
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.15584000945091248,
      "learning_rate": 0.0003920720646352128,
      "loss": 1.7547,
      "step": 7070
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.16131040453910828,
      "learning_rate": 0.0003920497716824596,
      "loss": 1.7804,
      "step": 7080
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1430855393409729,
      "learning_rate": 0.0003920274480659496,
      "loss": 1.7793,
      "step": 7090
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.15294766426086426,
      "learning_rate": 0.0003920050937892473,
      "loss": 1.7156,
      "step": 7100
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.13739879429340363,
      "learning_rate": 0.00039198270885592174,
      "loss": 1.7659,
      "step": 7110
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1557493805885315,
      "learning_rate": 0.0003919602932695472,
      "loss": 1.7918,
      "step": 7120
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.12877637147903442,
      "learning_rate": 0.00039193784703370264,
      "loss": 1.7088,
      "step": 7130
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.13669374585151672,
      "learning_rate": 0.00039191537015197185,
      "loss": 1.7093,
      "step": 7140
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.12947019934654236,
      "learning_rate": 0.00039189286262794376,
      "loss": 1.6805,
      "step": 7150
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.14283278584480286,
      "learning_rate": 0.0003918703244652119,
      "loss": 1.6415,
      "step": 7160
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.13801805675029755,
      "learning_rate": 0.00039184775566737494,
      "loss": 1.692,
      "step": 7170
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.16947269439697266,
      "learning_rate": 0.0003918251562380363,
      "loss": 1.7506,
      "step": 7180
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.13640892505645752,
      "learning_rate": 0.0003918025261808043,
      "loss": 1.6777,
      "step": 7190
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1325259953737259,
      "learning_rate": 0.0003917798654992923,
      "loss": 1.6815,
      "step": 7200
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.13371969759464264,
      "learning_rate": 0.00039175717419711833,
      "loss": 1.6929,
      "step": 7210
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.13861477375030518,
      "learning_rate": 0.0003917344522779054,
      "loss": 1.7506,
      "step": 7220
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.14447219669818878,
      "learning_rate": 0.00039171169974528157,
      "loss": 1.7032,
      "step": 7230
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.13237527012825012,
      "learning_rate": 0.0003916889166028795,
      "loss": 1.6761,
      "step": 7240
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1482444852590561,
      "learning_rate": 0.00039166610285433685,
      "loss": 1.711,
      "step": 7250
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.13787992298603058,
      "learning_rate": 0.0003916432585032963,
      "loss": 1.7039,
      "step": 7260
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.13764281570911407,
      "learning_rate": 0.0003916203835534052,
      "loss": 1.6459,
      "step": 7270
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.14277833700180054,
      "learning_rate": 0.000391597478008316,
      "loss": 1.7331,
      "step": 7280
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.14394375681877136,
      "learning_rate": 0.0003915745418716859,
      "loss": 1.7245,
      "step": 7290
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.12185429036617279,
      "learning_rate": 0.00039155157514717703,
      "loss": 1.6878,
      "step": 7300
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.12890471518039703,
      "learning_rate": 0.0003915285778384563,
      "loss": 1.6844,
      "step": 7310
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.13018891215324402,
      "learning_rate": 0.0003915055499491957,
      "loss": 1.6959,
      "step": 7320
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.169634148478508,
      "learning_rate": 0.0003914824914830719,
      "loss": 1.7361,
      "step": 7330
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1414484977722168,
      "learning_rate": 0.00039145940244376655,
      "loss": 1.7753,
      "step": 7340
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.14286017417907715,
      "learning_rate": 0.0003914362828349663,
      "loss": 1.6969,
      "step": 7350
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.13231833279132843,
      "learning_rate": 0.0003914131326603624,
      "loss": 1.7277,
      "step": 7360
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.14872434735298157,
      "learning_rate": 0.00039138995192365125,
      "loss": 1.736,
      "step": 7370
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1488579511642456,
      "learning_rate": 0.000391366740628534,
      "loss": 1.6918,
      "step": 7380
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1341143697500229,
      "learning_rate": 0.0003913434987787166,
      "loss": 1.6209,
      "step": 7390
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1347276121377945,
      "learning_rate": 0.0003913202263779101,
      "loss": 1.6827,
      "step": 7400
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.12921172380447388,
      "learning_rate": 0.00039129692342983023,
      "loss": 1.6474,
      "step": 7410
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1317102164030075,
      "learning_rate": 0.00039127358993819777,
      "loss": 1.6786,
      "step": 7420
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.14124788343906403,
      "learning_rate": 0.0003912502259067381,
      "loss": 1.764,
      "step": 7430
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.14556260406970978,
      "learning_rate": 0.0003912268313391818,
      "loss": 1.701,
      "step": 7440
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.13149979710578918,
      "learning_rate": 0.0003912034062392642,
      "loss": 1.7006,
      "step": 7450
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.13681216537952423,
      "learning_rate": 0.0003911799506107253,
      "loss": 1.706,
      "step": 7460
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.13194845616817474,
      "learning_rate": 0.0003911564644573103,
      "loss": 1.7716,
      "step": 7470
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.13681434094905853,
      "learning_rate": 0.0003911329477827692,
      "loss": 1.7356,
      "step": 7480
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1333349198102951,
      "learning_rate": 0.00039110940059085665,
      "loss": 1.7158,
      "step": 7490
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.13811297714710236,
      "learning_rate": 0.0003910858228853324,
      "loss": 1.7472,
      "step": 7500
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1425902247428894,
      "learning_rate": 0.000391062214669961,
      "loss": 1.7062,
      "step": 7510
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.13320201635360718,
      "learning_rate": 0.0003910385759485119,
      "loss": 1.6833,
      "step": 7520
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.13622677326202393,
      "learning_rate": 0.0003910149067247593,
      "loss": 1.7262,
      "step": 7530
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.13262036442756653,
      "learning_rate": 0.00039099120700248247,
      "loss": 1.68,
      "step": 7540
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.13869726657867432,
      "learning_rate": 0.00039096747678546537,
      "loss": 1.6351,
      "step": 7550
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.13586406409740448,
      "learning_rate": 0.000390943716077497,
      "loss": 1.7085,
      "step": 7560
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.13860741257667542,
      "learning_rate": 0.00039091992488237093,
      "loss": 1.7097,
      "step": 7570
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.14486007392406464,
      "learning_rate": 0.00039089610320388604,
      "loss": 1.711,
      "step": 7580
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.14704862236976624,
      "learning_rate": 0.00039087225104584563,
      "loss": 1.6913,
      "step": 7590
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.15811263024806976,
      "learning_rate": 0.0003908483684120582,
      "loss": 1.6907,
      "step": 7600
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.13361206650733948,
      "learning_rate": 0.0003908244553063369,
      "loss": 1.6782,
      "step": 7610
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.13869741559028625,
      "learning_rate": 0.0003908005117324999,
      "loss": 1.7114,
      "step": 7620
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.12840087711811066,
      "learning_rate": 0.0003907765376943702,
      "loss": 1.6481,
      "step": 7630
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.14002574980258942,
      "learning_rate": 0.0003907525331957755,
      "loss": 1.6603,
      "step": 7640
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.14109046757221222,
      "learning_rate": 0.00039072849824054866,
      "loss": 1.7,
      "step": 7650
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.12924519181251526,
      "learning_rate": 0.0003907044328325271,
      "loss": 1.6869,
      "step": 7660
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13750408589839935,
      "learning_rate": 0.00039068033697555333,
      "loss": 1.6731,
      "step": 7670
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.16169866919517517,
      "learning_rate": 0.0003906562106734745,
      "loss": 1.7461,
      "step": 7680
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13251982629299164,
      "learning_rate": 0.00039063205393014287,
      "loss": 1.6876,
      "step": 7690
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.15520523488521576,
      "learning_rate": 0.0003906078667494154,
      "loss": 1.7065,
      "step": 7700
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.12816987931728363,
      "learning_rate": 0.000390583649135154,
      "loss": 1.654,
      "step": 7710
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13748766481876373,
      "learning_rate": 0.00039055940109122535,
      "loss": 1.7374,
      "step": 7720
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.14492934942245483,
      "learning_rate": 0.0003905351226215011,
      "loss": 1.732,
      "step": 7730
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1277233511209488,
      "learning_rate": 0.0003905108137298575,
      "loss": 1.685,
      "step": 7740
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13751693069934845,
      "learning_rate": 0.00039048647442017605,
      "loss": 1.7166,
      "step": 7750
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2509312927722931,
      "learning_rate": 0.00039046210469634274,
      "loss": 2.0125,
      "step": 7760
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.14716836810112,
      "learning_rate": 0.00039043770456224876,
      "loss": 1.864,
      "step": 7770
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13424529135227203,
      "learning_rate": 0.00039041327402178984,
      "loss": 1.72,
      "step": 7780
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.12159378826618195,
      "learning_rate": 0.00039038881307886674,
      "loss": 1.7137,
      "step": 7790
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.12906280159950256,
      "learning_rate": 0.00039036432173738503,
      "loss": 1.7289,
      "step": 7800
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1405516117811203,
      "learning_rate": 0.00039033980000125515,
      "loss": 1.7146,
      "step": 7810
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1447688192129135,
      "learning_rate": 0.00039031524787439236,
      "loss": 1.7115,
      "step": 7820
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13648471236228943,
      "learning_rate": 0.00039029066536071683,
      "loss": 1.7023,
      "step": 7830
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1418386846780777,
      "learning_rate": 0.0003902660524641534,
      "loss": 1.7176,
      "step": 7840
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.12786869704723358,
      "learning_rate": 0.00039024140918863214,
      "loss": 1.6776,
      "step": 7850
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.12492866069078445,
      "learning_rate": 0.00039021673553808756,
      "loss": 1.661,
      "step": 7860
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.12730900943279266,
      "learning_rate": 0.0003901920315164592,
      "loss": 1.7203,
      "step": 7870
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13647036254405975,
      "learning_rate": 0.00039016729712769156,
      "loss": 1.7012,
      "step": 7880
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.14049872756004333,
      "learning_rate": 0.0003901425323757337,
      "loss": 1.6655,
      "step": 7890
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.11917974799871445,
      "learning_rate": 0.00039011773726453994,
      "loss": 1.6759,
      "step": 7900
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13427309691905975,
      "learning_rate": 0.000390092911798069,
      "loss": 1.6681,
      "step": 7910
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13851284980773926,
      "learning_rate": 0.00039006805598028473,
      "loss": 1.7163,
      "step": 7920
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1267043501138687,
      "learning_rate": 0.0003900431698151557,
      "loss": 1.7098,
      "step": 7930
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.22532662749290466,
      "learning_rate": 0.0003900182533066555,
      "loss": 1.6815,
      "step": 7940
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.14643071591854095,
      "learning_rate": 0.00038999330645876233,
      "loss": 1.7363,
      "step": 7950
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.15773718059062958,
      "learning_rate": 0.0003899683292754594,
      "loss": 1.7213,
      "step": 7960
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1535635143518448,
      "learning_rate": 0.00038994332176073466,
      "loss": 1.6854,
      "step": 7970
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.12380396574735641,
      "learning_rate": 0.00038991828391858103,
      "loss": 1.6882,
      "step": 7980
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13407306373119354,
      "learning_rate": 0.00038989321575299613,
      "loss": 1.7648,
      "step": 7990
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13147538900375366,
      "learning_rate": 0.00038986811726798246,
      "loss": 1.7537,
      "step": 8000
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13497044146060944,
      "learning_rate": 0.00038984298846754745,
      "loss": 1.6978,
      "step": 8010
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13441677391529083,
      "learning_rate": 0.0003898178293557033,
      "loss": 1.6786,
      "step": 8020
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.13331085443496704,
      "learning_rate": 0.000389792639936467,
      "loss": 1.6941,
      "step": 8030
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1468220353126526,
      "learning_rate": 0.0003897674202138605,
      "loss": 1.6199,
      "step": 8040
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1379866600036621,
      "learning_rate": 0.00038974217019191053,
      "loss": 1.7195,
      "step": 8050
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13483931124210358,
      "learning_rate": 0.0003897168898746486,
      "loss": 1.6968,
      "step": 8060
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13262777030467987,
      "learning_rate": 0.0003896915792661111,
      "loss": 1.6964,
      "step": 8070
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.12962651252746582,
      "learning_rate": 0.00038966623837033936,
      "loss": 1.7032,
      "step": 8080
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13349364697933197,
      "learning_rate": 0.0003896408671913793,
      "loss": 1.6942,
      "step": 8090
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.12977685034275055,
      "learning_rate": 0.000389615465733282,
      "loss": 1.7319,
      "step": 8100
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1472356915473938,
      "learning_rate": 0.0003895900340001031,
      "loss": 1.6924,
      "step": 8110
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1957908570766449,
      "learning_rate": 0.0003895645719959032,
      "loss": 1.6861,
      "step": 8120
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.14227932691574097,
      "learning_rate": 0.00038953907972474764,
      "loss": 1.7058,
      "step": 8130
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.12813404202461243,
      "learning_rate": 0.00038951355719070674,
      "loss": 1.6742,
      "step": 8140
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1380937248468399,
      "learning_rate": 0.00038948800439785557,
      "loss": 1.7324,
      "step": 8150
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13374464213848114,
      "learning_rate": 0.00038946242135027404,
      "loss": 1.6414,
      "step": 8160
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13308840990066528,
      "learning_rate": 0.0003894368080520468,
      "loss": 1.6614,
      "step": 8170
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13562729954719543,
      "learning_rate": 0.00038941116450726354,
      "loss": 1.6515,
      "step": 8180
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13918358087539673,
      "learning_rate": 0.0003893854907200185,
      "loss": 1.6598,
      "step": 8190
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.15010669827461243,
      "learning_rate": 0.00038935978669441104,
      "loss": 1.7217,
      "step": 8200
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1475742906332016,
      "learning_rate": 0.0003893340524345452,
      "loss": 1.6392,
      "step": 8210
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1437012404203415,
      "learning_rate": 0.00038930828794452976,
      "loss": 1.7013,
      "step": 8220
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.14719869196414948,
      "learning_rate": 0.00038928249322847853,
      "loss": 1.7308,
      "step": 8230
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.16744408011436462,
      "learning_rate": 0.00038925666829051,
      "loss": 1.7185,
      "step": 8240
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1367211937904358,
      "learning_rate": 0.0003892308131347475,
      "loss": 1.7045,
      "step": 8250
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13218046724796295,
      "learning_rate": 0.00038920492776531925,
      "loss": 1.6705,
      "step": 8260
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13375000655651093,
      "learning_rate": 0.0003891790121863582,
      "loss": 1.6399,
      "step": 8270
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13972359895706177,
      "learning_rate": 0.00038915306640200216,
      "loss": 1.6861,
      "step": 8280
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13885724544525146,
      "learning_rate": 0.00038912709041639395,
      "loss": 1.6812,
      "step": 8290
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.14410728216171265,
      "learning_rate": 0.0003891010842336809,
      "loss": 1.7016,
      "step": 8300
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13372208178043365,
      "learning_rate": 0.0003890750478580153,
      "loss": 1.6539,
      "step": 8310
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.133956640958786,
      "learning_rate": 0.00038904898129355435,
      "loss": 1.6481,
      "step": 8320
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.14632944762706757,
      "learning_rate": 0.00038902288454445997,
      "loss": 1.6756,
      "step": 8330
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1364748626947403,
      "learning_rate": 0.0003889967576148988,
      "loss": 1.6624,
      "step": 8340
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13148286938667297,
      "learning_rate": 0.0003889706005090425,
      "loss": 1.6931,
      "step": 8350
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.14017271995544434,
      "learning_rate": 0.0003889444132310675,
      "loss": 1.7128,
      "step": 8360
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1335548460483551,
      "learning_rate": 0.00038891819578515494,
      "loss": 1.7638,
      "step": 8370
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13582909107208252,
      "learning_rate": 0.00038889194817549085,
      "loss": 1.6896,
      "step": 8380
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1565767079591751,
      "learning_rate": 0.00038886567040626616,
      "loss": 1.7231,
      "step": 8390
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.15570221841335297,
      "learning_rate": 0.0003888393624816764,
      "loss": 1.6965,
      "step": 8400
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.14114753901958466,
      "learning_rate": 0.0003888130244059221,
      "loss": 1.6967,
      "step": 8410
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13201847672462463,
      "learning_rate": 0.00038878665618320864,
      "loss": 1.6502,
      "step": 8420
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13043740391731262,
      "learning_rate": 0.00038876025781774603,
      "loss": 1.7026,
      "step": 8430
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.16153882443904877,
      "learning_rate": 0.0003887338293137491,
      "loss": 1.6899,
      "step": 8440
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.12798282504081726,
      "learning_rate": 0.0003887073706754377,
      "loss": 1.6431,
      "step": 8450
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.13147814571857452,
      "learning_rate": 0.0003886808819070363,
      "loss": 1.6663,
      "step": 8460
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.13304992020130157,
      "learning_rate": 0.0003886543630127743,
      "loss": 1.6965,
      "step": 8470
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.13692429661750793,
      "learning_rate": 0.00038862781399688585,
      "loss": 1.6628,
      "step": 8480
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.22576695680618286,
      "learning_rate": 0.00038860123486360987,
      "loss": 1.6373,
      "step": 8490
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.15108773112297058,
      "learning_rate": 0.0003885746256171902,
      "loss": 1.655,
      "step": 8500
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1294201761484146,
      "learning_rate": 0.0003885479862618754,
      "loss": 1.6525,
      "step": 8510
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.12410447001457214,
      "learning_rate": 0.00038852131680191875,
      "loss": 1.6522,
      "step": 8520
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1444026529788971,
      "learning_rate": 0.0003884946172415786,
      "loss": 1.6696,
      "step": 8530
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.134349063038826,
      "learning_rate": 0.00038846788758511785,
      "loss": 1.6533,
      "step": 8540
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.12954683601856232,
      "learning_rate": 0.00038844112783680446,
      "loss": 1.6767,
      "step": 8550
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.13951031863689423,
      "learning_rate": 0.00038841433800091085,
      "loss": 1.7252,
      "step": 8560
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.14629290997982025,
      "learning_rate": 0.0003883875180817146,
      "loss": 1.6573,
      "step": 8570
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.15033113956451416,
      "learning_rate": 0.00038836066808349784,
      "loss": 1.7045,
      "step": 8580
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.13196027278900146,
      "learning_rate": 0.00038833378801054753,
      "loss": 1.7253,
      "step": 8590
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1399390995502472,
      "learning_rate": 0.00038830687786715564,
      "loss": 1.6248,
      "step": 8600
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.12518121302127838,
      "learning_rate": 0.00038827993765761876,
      "loss": 1.6735,
      "step": 8610
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.13088391721248627,
      "learning_rate": 0.00038825296738623824,
      "loss": 1.6529,
      "step": 8620
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1249965950846672,
      "learning_rate": 0.00038822596705732036,
      "loss": 1.7086,
      "step": 8630
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.13543793559074402,
      "learning_rate": 0.00038819893667517614,
      "loss": 1.6193,
      "step": 8640
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.16318467259407043,
      "learning_rate": 0.00038817187624412146,
      "loss": 1.7019,
      "step": 8650
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1324823647737503,
      "learning_rate": 0.00038814478576847686,
      "loss": 1.7179,
      "step": 8660
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1390790194272995,
      "learning_rate": 0.0003881176652525677,
      "loss": 1.6679,
      "step": 8670
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.12248550355434418,
      "learning_rate": 0.00038809051470072443,
      "loss": 1.6866,
      "step": 8680
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.14330117404460907,
      "learning_rate": 0.0003880633341172818,
      "loss": 1.6583,
      "step": 8690
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.12497661262750626,
      "learning_rate": 0.00038803612350657977,
      "loss": 1.6371,
      "step": 8700
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.14796240627765656,
      "learning_rate": 0.0003880088828729629,
      "loss": 1.669,
      "step": 8710
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.198830246925354,
      "learning_rate": 0.00038798161222078055,
      "loss": 1.6904,
      "step": 8720
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1437385082244873,
      "learning_rate": 0.00038795431155438703,
      "loss": 1.6763,
      "step": 8730
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1363937258720398,
      "learning_rate": 0.00038792698087814116,
      "loss": 1.6551,
      "step": 8740
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.13170404732227325,
      "learning_rate": 0.0003878996201964068,
      "loss": 1.6615,
      "step": 8750
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.12872740626335144,
      "learning_rate": 0.00038787222951355247,
      "loss": 1.6544,
      "step": 8760
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.15284587442874908,
      "learning_rate": 0.0003878448088339516,
      "loss": 1.6436,
      "step": 8770
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.14942800998687744,
      "learning_rate": 0.0003878173581619822,
      "loss": 1.6953,
      "step": 8780
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1378229409456253,
      "learning_rate": 0.00038778987750202735,
      "loss": 1.7536,
      "step": 8790
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.13812203705310822,
      "learning_rate": 0.0003877623668584747,
      "loss": 1.7113,
      "step": 8800
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1441105157136917,
      "learning_rate": 0.0003877348262357167,
      "loss": 1.7004,
      "step": 8810
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1562652587890625,
      "learning_rate": 0.0003877072556381507,
      "loss": 1.7129,
      "step": 8820
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.13454972207546234,
      "learning_rate": 0.0003876796550701788,
      "loss": 1.6665,
      "step": 8830
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.14589405059814453,
      "learning_rate": 0.00038765202453620775,
      "loss": 1.6646,
      "step": 8840
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13198499381542206,
      "learning_rate": 0.0003876243640406494,
      "loss": 1.6491,
      "step": 8850
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.16170230507850647,
      "learning_rate": 0.00038759667358792,
      "loss": 1.6992,
      "step": 8860
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.12688037753105164,
      "learning_rate": 0.00038756895318244083,
      "loss": 1.651,
      "step": 8870
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1396840214729309,
      "learning_rate": 0.0003875412028286379,
      "loss": 1.6469,
      "step": 8880
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13911014795303345,
      "learning_rate": 0.00038751342253094196,
      "loss": 1.6992,
      "step": 8890
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1247856467962265,
      "learning_rate": 0.0003874856122937886,
      "loss": 1.6641,
      "step": 8900
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1389637291431427,
      "learning_rate": 0.0003874577721216182,
      "loss": 1.66,
      "step": 8910
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.11789216846227646,
      "learning_rate": 0.0003874299020188757,
      "loss": 1.6096,
      "step": 8920
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1422717571258545,
      "learning_rate": 0.0003874020019900112,
      "loss": 1.6885,
      "step": 8930
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13758403062820435,
      "learning_rate": 0.0003873740720394793,
      "loss": 1.6719,
      "step": 8940
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1851162314414978,
      "learning_rate": 0.00038734611217173945,
      "loss": 1.841,
      "step": 8950
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.16305282711982727,
      "learning_rate": 0.00038731812239125587,
      "loss": 1.6986,
      "step": 8960
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1490446776151657,
      "learning_rate": 0.00038729010270249764,
      "loss": 1.7077,
      "step": 8970
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13517969846725464,
      "learning_rate": 0.0003872620531099384,
      "loss": 1.6802,
      "step": 8980
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13483189046382904,
      "learning_rate": 0.00038723397361805685,
      "loss": 1.6995,
      "step": 8990
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13785387575626373,
      "learning_rate": 0.0003872058642313363,
      "loss": 1.7246,
      "step": 9000
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.127422034740448,
      "learning_rate": 0.00038717772495426475,
      "loss": 1.6759,
      "step": 9010
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.14029677212238312,
      "learning_rate": 0.00038714955579133514,
      "loss": 1.691,
      "step": 9020
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1330009549856186,
      "learning_rate": 0.0003871213567470452,
      "loss": 1.6435,
      "step": 9030
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.12364694476127625,
      "learning_rate": 0.0003870931278258972,
      "loss": 1.6694,
      "step": 9040
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13656748831272125,
      "learning_rate": 0.0003870648690323984,
      "loss": 1.7141,
      "step": 9050
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1401910036802292,
      "learning_rate": 0.00038703658037106083,
      "loss": 1.6992,
      "step": 9060
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13535091280937195,
      "learning_rate": 0.0003870082618464011,
      "loss": 1.6243,
      "step": 9070
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13311734795570374,
      "learning_rate": 0.0003869799134629408,
      "loss": 1.6122,
      "step": 9080
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.12910255789756775,
      "learning_rate": 0.0003869515352252061,
      "loss": 1.7029,
      "step": 9090
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.12881791591644287,
      "learning_rate": 0.0003869231271377281,
      "loss": 1.6627,
      "step": 9100
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.12921388447284698,
      "learning_rate": 0.0003868946892050426,
      "loss": 1.6297,
      "step": 9110
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.12428681552410126,
      "learning_rate": 0.0003868662214316901,
      "loss": 1.6419,
      "step": 9120
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.15118862688541412,
      "learning_rate": 0.000386837723822216,
      "loss": 1.7649,
      "step": 9130
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1526978313922882,
      "learning_rate": 0.00038680919638117033,
      "loss": 1.6654,
      "step": 9140
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.15959547460079193,
      "learning_rate": 0.00038678063911310796,
      "loss": 1.6896,
      "step": 9150
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.14070424437522888,
      "learning_rate": 0.0003867520520225886,
      "loss": 1.6588,
      "step": 9160
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.15812811255455017,
      "learning_rate": 0.00038672343511417646,
      "loss": 1.6973,
      "step": 9170
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1548108160495758,
      "learning_rate": 0.0003866947883924408,
      "loss": 1.6617,
      "step": 9180
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1341230422258377,
      "learning_rate": 0.0003866661118619554,
      "loss": 1.7306,
      "step": 9190
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13387803733348846,
      "learning_rate": 0.0003866374055272991,
      "loss": 1.6905,
      "step": 9200
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.17781074345111847,
      "learning_rate": 0.0003866086693930552,
      "loss": 1.6869,
      "step": 9210
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13148914277553558,
      "learning_rate": 0.0003865799034638118,
      "loss": 1.6822,
      "step": 9220
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.12276733666658401,
      "learning_rate": 0.00038655110774416196,
      "loss": 1.7114,
      "step": 9230
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1315593272447586,
      "learning_rate": 0.00038652228223870337,
      "loss": 1.683,
      "step": 9240
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.13443982601165771,
      "learning_rate": 0.0003864934269520384,
      "loss": 1.6736,
      "step": 9250
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.132383331656456,
      "learning_rate": 0.0003864645418887743,
      "loss": 1.6831,
      "step": 9260
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.12144575268030167,
      "learning_rate": 0.00038643562705352295,
      "loss": 1.6697,
      "step": 9270
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.14845208823680878,
      "learning_rate": 0.0003864066824509012,
      "loss": 1.6565,
      "step": 9280
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.13534559309482574,
      "learning_rate": 0.0003863777080855303,
      "loss": 1.6132,
      "step": 9290
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.12392674386501312,
      "learning_rate": 0.0003863487039620366,
      "loss": 1.7073,
      "step": 9300
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.12857109308242798,
      "learning_rate": 0.0003863196700850511,
      "loss": 1.6741,
      "step": 9310
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.13830944895744324,
      "learning_rate": 0.0003862906064592094,
      "loss": 1.6617,
      "step": 9320
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.13779057562351227,
      "learning_rate": 0.0003862615130891521,
      "loss": 1.6576,
      "step": 9330
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.12795256078243256,
      "learning_rate": 0.0003862323899795243,
      "loss": 1.6418,
      "step": 9340
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.12998293340206146,
      "learning_rate": 0.000386203237134976,
      "loss": 1.645,
      "step": 9350
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.13535144925117493,
      "learning_rate": 0.00038617405456016187,
      "loss": 1.7063,
      "step": 9360
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.14552073180675507,
      "learning_rate": 0.00038614484225974144,
      "loss": 1.7002,
      "step": 9370
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.13565002381801605,
      "learning_rate": 0.00038611560023837883,
      "loss": 1.6609,
      "step": 9380
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.13093070685863495,
      "learning_rate": 0.00038608632850074305,
      "loss": 1.6438,
      "step": 9390
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.15767642855644226,
      "learning_rate": 0.0003860570270515077,
      "loss": 1.7212,
      "step": 9400
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.12853416800498962,
      "learning_rate": 0.0003860276958953514,
      "loss": 1.6376,
      "step": 9410
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.15520334243774414,
      "learning_rate": 0.00038599833503695713,
      "loss": 1.7403,
      "step": 9420
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.13134929537773132,
      "learning_rate": 0.00038596894448101297,
      "loss": 1.6265,
      "step": 9430
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.12647289037704468,
      "learning_rate": 0.0003859395242322115,
      "loss": 1.6685,
      "step": 9440
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1194876879453659,
      "learning_rate": 0.00038591007429525,
      "loss": 1.629,
      "step": 9450
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.12292616069316864,
      "learning_rate": 0.0003858805946748309,
      "loss": 1.6883,
      "step": 9460
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1364501565694809,
      "learning_rate": 0.00038585108537566085,
      "loss": 1.6677,
      "step": 9470
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1332101970911026,
      "learning_rate": 0.00038582154640245156,
      "loss": 1.6828,
      "step": 9480
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.13498814404010773,
      "learning_rate": 0.0003857919777599194,
      "loss": 1.6912,
      "step": 9490
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1320127248764038,
      "learning_rate": 0.00038576237945278543,
      "loss": 1.6613,
      "step": 9500
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.12627199292182922,
      "learning_rate": 0.0003857327514857755,
      "loss": 1.6771,
      "step": 9510
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.12172795087099075,
      "learning_rate": 0.00038570309386362015,
      "loss": 1.6874,
      "step": 9520
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.14356915652751923,
      "learning_rate": 0.00038567340659105483,
      "loss": 1.7213,
      "step": 9530
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.14241376519203186,
      "learning_rate": 0.00038564368967281936,
      "loss": 1.7341,
      "step": 9540
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.15648552775382996,
      "learning_rate": 0.00038561394311365866,
      "loss": 1.7055,
      "step": 9550
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1445528119802475,
      "learning_rate": 0.00038558416691832217,
      "loss": 1.6674,
      "step": 9560
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.13047567009925842,
      "learning_rate": 0.0003855543610915642,
      "loss": 1.6434,
      "step": 9570
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.137510746717453,
      "learning_rate": 0.0003855245256381436,
      "loss": 1.6924,
      "step": 9580
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.12807559967041016,
      "learning_rate": 0.00038549466056282417,
      "loss": 1.6805,
      "step": 9590
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.13783133029937744,
      "learning_rate": 0.0003854647658703743,
      "loss": 1.6968,
      "step": 9600
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.13350753486156464,
      "learning_rate": 0.0003854348415655671,
      "loss": 1.6599,
      "step": 9610
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.15386106073856354,
      "learning_rate": 0.0003854048876531805,
      "loss": 1.6706,
      "step": 9620
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.14253292977809906,
      "learning_rate": 0.0003853749041379972,
      "loss": 1.6567,
      "step": 9630
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1428574025630951,
      "learning_rate": 0.00038534489102480437,
      "loss": 1.6911,
      "step": 9640
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.12866978347301483,
      "learning_rate": 0.0003853148483183942,
      "loss": 1.6676,
      "step": 9650
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.12328783422708511,
      "learning_rate": 0.00038528477602356345,
      "loss": 1.639,
      "step": 9660
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.12650468945503235,
      "learning_rate": 0.00038525467414511356,
      "loss": 1.6606,
      "step": 9670
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.13679854571819305,
      "learning_rate": 0.0003852245426878508,
      "loss": 1.6582,
      "step": 9680
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.13160301744937897,
      "learning_rate": 0.0003851943816565862,
      "loss": 1.6697,
      "step": 9690
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.13433463871479034,
      "learning_rate": 0.00038516419105613544,
      "loss": 1.5862,
      "step": 9700
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.12435674667358398,
      "learning_rate": 0.0003851339708913188,
      "loss": 1.6992,
      "step": 9710
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1282668113708496,
      "learning_rate": 0.0003851037211669615,
      "loss": 1.6415,
      "step": 9720
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.13300156593322754,
      "learning_rate": 0.00038507344188789335,
      "loss": 1.6636,
      "step": 9730
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.13623046875,
      "learning_rate": 0.00038504313305894896,
      "loss": 1.6138,
      "step": 9740
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1425723135471344,
      "learning_rate": 0.0003850127946849676,
      "loss": 1.6753,
      "step": 9750
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1205907091498375,
      "learning_rate": 0.0003849824267707932,
      "loss": 1.6981,
      "step": 9760
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.13399383425712585,
      "learning_rate": 0.0003849520293212745,
      "loss": 1.6687,
      "step": 9770
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.14110614359378815,
      "learning_rate": 0.000384921602341265,
      "loss": 1.6718,
      "step": 9780
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.14635838568210602,
      "learning_rate": 0.0003848911458356227,
      "loss": 1.6424,
      "step": 9790
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.12122263759374619,
      "learning_rate": 0.0003848606598092106,
      "loss": 1.6415,
      "step": 9800
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.13565990328788757,
      "learning_rate": 0.00038483014426689624,
      "loss": 1.6716,
      "step": 9810
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1413806825876236,
      "learning_rate": 0.0003847995992135519,
      "loss": 1.6768,
      "step": 9820
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.14945995807647705,
      "learning_rate": 0.0003847690246540545,
      "loss": 1.6907,
      "step": 9830
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1399151086807251,
      "learning_rate": 0.00038473842059328587,
      "loss": 1.7095,
      "step": 9840
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1341298669576645,
      "learning_rate": 0.00038470778703613226,
      "loss": 1.6107,
      "step": 9850
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1368744671344757,
      "learning_rate": 0.00038467712398748504,
      "loss": 1.6942,
      "step": 9860
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.12841053307056427,
      "learning_rate": 0.00038464643145223976,
      "loss": 1.6767,
      "step": 9870
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.12721240520477295,
      "learning_rate": 0.0003846157094352973,
      "loss": 1.6485,
      "step": 9880
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1607305407524109,
      "learning_rate": 0.0003845849579415626,
      "loss": 1.6569,
      "step": 9890
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.12862354516983032,
      "learning_rate": 0.00038455417697594573,
      "loss": 1.6933,
      "step": 9900
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.12964557111263275,
      "learning_rate": 0.0003845233665433614,
      "loss": 1.6559,
      "step": 9910
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.13759677112102509,
      "learning_rate": 0.00038449252664872897,
      "loss": 1.6401,
      "step": 9920
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.13009068369865417,
      "learning_rate": 0.00038446165729697244,
      "loss": 1.7112,
      "step": 9930
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.12266763299703598,
      "learning_rate": 0.0003844307584930207,
      "loss": 1.6665,
      "step": 9940
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.16423115134239197,
      "learning_rate": 0.0003843998302418071,
      "loss": 1.6358,
      "step": 9950
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1375732421875,
      "learning_rate": 0.0003843688725482699,
      "loss": 1.643,
      "step": 9960
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1389283388853073,
      "learning_rate": 0.000384337885417352,
      "loss": 1.7376,
      "step": 9970
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.14045849442481995,
      "learning_rate": 0.00038430686885400096,
      "loss": 1.6542,
      "step": 9980
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1282881647348404,
      "learning_rate": 0.000384275822863169,
      "loss": 1.6977,
      "step": 9990
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.14488451182842255,
      "learning_rate": 0.00038424474744981315,
      "loss": 1.6779,
      "step": 10000
    }
  ],
  "logging_steps": 10,
  "max_steps": 78622,
  "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
}