{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.968781408006432,
  "global_step": 24000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 2.0610057708161583e-07,
      "loss": 4.8665,
      "step": 10
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.1220115416323167e-07,
      "loss": 5.037,
      "step": 20
    },
    {
      "epoch": 0.0,
      "learning_rate": 6.183017312448475e-07,
      "loss": 4.9416,
      "step": 30
    },
    {
      "epoch": 0.0,
      "learning_rate": 8.244023083264633e-07,
      "loss": 4.8036,
      "step": 40
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.0305028854080792e-06,
      "loss": 4.7498,
      "step": 50
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.236603462489695e-06,
      "loss": 4.7409,
      "step": 60
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.442704039571311e-06,
      "loss": 4.5851,
      "step": 70
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.6488046166529267e-06,
      "loss": 4.3809,
      "step": 80
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.8549051937345424e-06,
      "loss": 4.0034,
      "step": 90
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.0610057708161584e-06,
      "loss": 4.0203,
      "step": 100
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.267106347897774e-06,
      "loss": 3.8694,
      "step": 110
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.47320692497939e-06,
      "loss": 3.7423,
      "step": 120
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.679307502061006e-06,
      "loss": 3.6758,
      "step": 130
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.885408079142622e-06,
      "loss": 3.5005,
      "step": 140
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.0915086562242376e-06,
      "loss": 3.4159,
      "step": 150
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.2976092333058533e-06,
      "loss": 3.2655,
      "step": 160
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.503709810387469e-06,
      "loss": 3.2183,
      "step": 170
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.709810387469085e-06,
      "loss": 3.0816,
      "step": 180
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.915910964550701e-06,
      "loss": 2.9692,
      "step": 190
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.122011541632317e-06,
      "loss": 2.8064,
      "step": 200
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.328112118713933e-06,
      "loss": 2.7355,
      "step": 210
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.534212695795548e-06,
      "loss": 2.5492,
      "step": 220
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.7403132728771645e-06,
      "loss": 2.5349,
      "step": 230
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.94641384995878e-06,
      "loss": 2.5254,
      "step": 240
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.152514427040396e-06,
      "loss": 2.4401,
      "step": 250
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.358615004122012e-06,
      "loss": 2.2764,
      "step": 260
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.5647155812036275e-06,
      "loss": 2.2592,
      "step": 270
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.770816158285244e-06,
      "loss": 2.1613,
      "step": 280
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.97691673536686e-06,
      "loss": 2.0636,
      "step": 290
    },
    {
      "epoch": 0.04,
      "learning_rate": 6.183017312448475e-06,
      "loss": 2.0125,
      "step": 300
    },
    {
      "epoch": 0.04,
      "learning_rate": 6.389117889530091e-06,
      "loss": 1.9533,
      "step": 310
    },
    {
      "epoch": 0.04,
      "learning_rate": 6.595218466611707e-06,
      "loss": 1.9327,
      "step": 320
    },
    {
      "epoch": 0.04,
      "learning_rate": 6.801319043693323e-06,
      "loss": 1.8861,
      "step": 330
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.007419620774938e-06,
      "loss": 1.8622,
      "step": 340
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.213520197856554e-06,
      "loss": 1.7515,
      "step": 350
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.41962077493817e-06,
      "loss": 1.7603,
      "step": 360
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.625721352019787e-06,
      "loss": 1.6659,
      "step": 370
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.831821929101401e-06,
      "loss": 1.6916,
      "step": 380
    },
    {
      "epoch": 0.05,
      "learning_rate": 8.037922506183018e-06,
      "loss": 1.7078,
      "step": 390
    },
    {
      "epoch": 0.05,
      "learning_rate": 8.244023083264634e-06,
      "loss": 1.6047,
      "step": 400
    },
    {
      "epoch": 0.05,
      "learning_rate": 8.45012366034625e-06,
      "loss": 1.5654,
      "step": 410
    },
    {
      "epoch": 0.05,
      "learning_rate": 8.656224237427866e-06,
      "loss": 1.5927,
      "step": 420
    },
    {
      "epoch": 0.05,
      "learning_rate": 8.862324814509481e-06,
      "loss": 1.6095,
      "step": 430
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.068425391591097e-06,
      "loss": 1.6156,
      "step": 440
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.274525968672712e-06,
      "loss": 1.5368,
      "step": 450
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.480626545754329e-06,
      "loss": 1.5421,
      "step": 460
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.686727122835944e-06,
      "loss": 1.5215,
      "step": 470
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.89282769991756e-06,
      "loss": 1.5233,
      "step": 480
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.0098928276999175e-05,
      "loss": 1.4729,
      "step": 490
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.0305028854080792e-05,
      "loss": 1.539,
      "step": 500
    },
    {
      "epoch": 0.06,
      "eval_loss": 1.3461134433746338,
      "eval_runtime": 5.768,
      "eval_samples_per_second": 89.806,
      "eval_steps_per_second": 11.269,
      "step": 500
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.0511129431162407e-05,
      "loss": 1.5064,
      "step": 510
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.0717230008244024e-05,
      "loss": 1.5028,
      "step": 520
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.092333058532564e-05,
      "loss": 1.4774,
      "step": 530
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.1129431162407255e-05,
      "loss": 1.4405,
      "step": 540
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.133553173948887e-05,
      "loss": 1.4486,
      "step": 550
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.1541632316570487e-05,
      "loss": 1.4701,
      "step": 560
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.1747732893652103e-05,
      "loss": 1.4776,
      "step": 570
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.195383347073372e-05,
      "loss": 1.4349,
      "step": 580
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.2159934047815335e-05,
      "loss": 1.4336,
      "step": 590
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.236603462489695e-05,
      "loss": 1.4612,
      "step": 600
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.2572135201978566e-05,
      "loss": 1.4073,
      "step": 610
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.2778235779060183e-05,
      "loss": 1.3991,
      "step": 620
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.29843363561418e-05,
      "loss": 1.4243,
      "step": 630
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.3190436933223413e-05,
      "loss": 1.4053,
      "step": 640
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.3396537510305029e-05,
      "loss": 1.4047,
      "step": 650
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.3602638087386646e-05,
      "loss": 1.3824,
      "step": 660
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.380873866446826e-05,
      "loss": 1.3764,
      "step": 670
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.4014839241549876e-05,
      "loss": 1.4373,
      "step": 680
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.4220939818631493e-05,
      "loss": 1.4134,
      "step": 690
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.4427040395713109e-05,
      "loss": 1.3613,
      "step": 700
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.4633140972794724e-05,
      "loss": 1.416,
      "step": 710
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.483924154987634e-05,
      "loss": 1.4148,
      "step": 720
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.5045342126957956e-05,
      "loss": 1.3494,
      "step": 730
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.5251442704039573e-05,
      "loss": 1.3827,
      "step": 740
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.5457543281121185e-05,
      "loss": 1.3406,
      "step": 750
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.5663643858202802e-05,
      "loss": 1.404,
      "step": 760
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.586974443528442e-05,
      "loss": 1.3474,
      "step": 770
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.6075845012366037e-05,
      "loss": 1.3925,
      "step": 780
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.628194558944765e-05,
      "loss": 1.3944,
      "step": 790
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.6488046166529267e-05,
      "loss": 1.3822,
      "step": 800
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.6694146743610884e-05,
      "loss": 1.3885,
      "step": 810
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.69002473206925e-05,
      "loss": 1.3641,
      "step": 820
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.7106347897774115e-05,
      "loss": 1.4021,
      "step": 830
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.7312448474855732e-05,
      "loss": 1.3464,
      "step": 840
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.7518549051937346e-05,
      "loss": 1.3972,
      "step": 850
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.7724649629018963e-05,
      "loss": 1.3618,
      "step": 860
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.7930750206100576e-05,
      "loss": 1.4035,
      "step": 870
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.8136850783182193e-05,
      "loss": 1.3637,
      "step": 880
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.834295136026381e-05,
      "loss": 1.4274,
      "step": 890
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.8549051937345424e-05,
      "loss": 1.3806,
      "step": 900
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.875515251442704e-05,
      "loss": 1.3702,
      "step": 910
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.8961253091508658e-05,
      "loss": 1.3692,
      "step": 920
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9167353668590275e-05,
      "loss": 1.3833,
      "step": 930
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.937345424567189e-05,
      "loss": 1.3583,
      "step": 940
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9579554822753506e-05,
      "loss": 1.3478,
      "step": 950
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.978565539983512e-05,
      "loss": 1.3563,
      "step": 960
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9991755976916736e-05,
      "loss": 1.3484,
      "step": 970
    },
    {
      "epoch": 0.12,
      "learning_rate": 2.019785655399835e-05,
      "loss": 1.3945,
      "step": 980
    },
    {
      "epoch": 0.12,
      "learning_rate": 2.0403957131079967e-05,
      "loss": 1.3447,
      "step": 990
    },
    {
      "epoch": 0.12,
      "learning_rate": 2.0610057708161584e-05,
      "loss": 1.3477,
      "step": 1000
    },
    {
      "epoch": 0.12,
      "eval_loss": 1.2597277164459229,
      "eval_runtime": 5.7653,
      "eval_samples_per_second": 89.848,
      "eval_steps_per_second": 11.274,
      "step": 1000
    },
    {
      "epoch": 0.12,
      "learning_rate": 2.08161582852432e-05,
      "loss": 1.3636,
      "step": 1010
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.1022258862324815e-05,
      "loss": 1.3576,
      "step": 1020
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.122835943940643e-05,
      "loss": 1.3677,
      "step": 1030
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.143446001648805e-05,
      "loss": 1.337,
      "step": 1040
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.1640560593569666e-05,
      "loss": 1.3743,
      "step": 1050
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.184666117065128e-05,
      "loss": 1.3512,
      "step": 1060
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.2052761747732893e-05,
      "loss": 1.3447,
      "step": 1070
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.225886232481451e-05,
      "loss": 1.3772,
      "step": 1080
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.2464962901896127e-05,
      "loss": 1.3257,
      "step": 1090
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.267106347897774e-05,
      "loss": 1.3911,
      "step": 1100
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.2877164056059358e-05,
      "loss": 1.3638,
      "step": 1110
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.3083264633140975e-05,
      "loss": 1.3602,
      "step": 1120
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.328936521022259e-05,
      "loss": 1.344,
      "step": 1130
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.3495465787304205e-05,
      "loss": 1.3768,
      "step": 1140
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.3701566364385822e-05,
      "loss": 1.3314,
      "step": 1150
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.390766694146744e-05,
      "loss": 1.3609,
      "step": 1160
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.4113767518549053e-05,
      "loss": 1.3948,
      "step": 1170
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.431986809563067e-05,
      "loss": 1.3339,
      "step": 1180
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.4525968672712284e-05,
      "loss": 1.3364,
      "step": 1190
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.47320692497939e-05,
      "loss": 1.3286,
      "step": 1200
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.4938169826875514e-05,
      "loss": 1.3242,
      "step": 1210
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.514427040395713e-05,
      "loss": 1.3238,
      "step": 1220
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.5350370981038745e-05,
      "loss": 1.3601,
      "step": 1230
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.5556471558120365e-05,
      "loss": 1.3517,
      "step": 1240
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.576257213520198e-05,
      "loss": 1.3391,
      "step": 1250
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.59686727122836e-05,
      "loss": 1.3292,
      "step": 1260
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.6174773289365213e-05,
      "loss": 1.3411,
      "step": 1270
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.6380873866446827e-05,
      "loss": 1.3241,
      "step": 1280
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.6586974443528444e-05,
      "loss": 1.3354,
      "step": 1290
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.6793075020610057e-05,
      "loss": 1.3345,
      "step": 1300
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.699917559769167e-05,
      "loss": 1.3384,
      "step": 1310
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.720527617477329e-05,
      "loss": 1.3396,
      "step": 1320
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.7411376751854905e-05,
      "loss": 1.3133,
      "step": 1330
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.761747732893652e-05,
      "loss": 1.328,
      "step": 1340
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.782357790601814e-05,
      "loss": 1.3167,
      "step": 1350
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.8029678483099753e-05,
      "loss": 1.3062,
      "step": 1360
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.8235779060181373e-05,
      "loss": 1.2965,
      "step": 1370
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.8441879637262987e-05,
      "loss": 1.3645,
      "step": 1380
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.86479802143446e-05,
      "loss": 1.3682,
      "step": 1390
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.8854080791426218e-05,
      "loss": 1.3035,
      "step": 1400
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.906018136850783e-05,
      "loss": 1.3153,
      "step": 1410
    },
    {
      "epoch": 0.18,
      "learning_rate": 2.9266281945589448e-05,
      "loss": 1.3141,
      "step": 1420
    },
    {
      "epoch": 0.18,
      "learning_rate": 2.9472382522671065e-05,
      "loss": 1.3191,
      "step": 1430
    },
    {
      "epoch": 0.18,
      "learning_rate": 2.967848309975268e-05,
      "loss": 1.3252,
      "step": 1440
    },
    {
      "epoch": 0.18,
      "learning_rate": 2.98845836768343e-05,
      "loss": 1.3211,
      "step": 1450
    },
    {
      "epoch": 0.18,
      "learning_rate": 3.0090684253915913e-05,
      "loss": 1.3037,
      "step": 1460
    },
    {
      "epoch": 0.18,
      "learning_rate": 3.0296784830997527e-05,
      "loss": 1.3206,
      "step": 1470
    },
    {
      "epoch": 0.18,
      "learning_rate": 3.0502885408079147e-05,
      "loss": 1.2752,
      "step": 1480
    },
    {
      "epoch": 0.18,
      "learning_rate": 3.070898598516076e-05,
      "loss": 1.3213,
      "step": 1490
    },
    {
      "epoch": 0.19,
      "learning_rate": 3.091508656224237e-05,
      "loss": 1.3303,
      "step": 1500
    },
    {
      "epoch": 0.19,
      "eval_loss": 1.2215049266815186,
      "eval_runtime": 5.7676,
      "eval_samples_per_second": 89.812,
      "eval_steps_per_second": 11.27,
      "step": 1500
    },
    {
      "epoch": 0.19,
      "learning_rate": 3.112118713932399e-05,
      "loss": 1.3255,
      "step": 1510
    },
    {
      "epoch": 0.19,
      "learning_rate": 3.1327287716405605e-05,
      "loss": 1.3062,
      "step": 1520
    },
    {
      "epoch": 0.19,
      "learning_rate": 3.1533388293487225e-05,
      "loss": 1.2983,
      "step": 1530
    },
    {
      "epoch": 0.19,
      "learning_rate": 3.173948887056884e-05,
      "loss": 1.3201,
      "step": 1540
    },
    {
      "epoch": 0.19,
      "learning_rate": 3.194558944765045e-05,
      "loss": 1.2797,
      "step": 1550
    },
    {
      "epoch": 0.19,
      "learning_rate": 3.215169002473207e-05,
      "loss": 1.3081,
      "step": 1560
    },
    {
      "epoch": 0.19,
      "learning_rate": 3.235779060181369e-05,
      "loss": 1.3323,
      "step": 1570
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.25638911788953e-05,
      "loss": 1.3209,
      "step": 1580
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.276999175597692e-05,
      "loss": 1.3459,
      "step": 1590
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.2976092333058534e-05,
      "loss": 1.3325,
      "step": 1600
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.318219291014015e-05,
      "loss": 1.3048,
      "step": 1610
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.338829348722177e-05,
      "loss": 1.2714,
      "step": 1620
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.359439406430338e-05,
      "loss": 1.2817,
      "step": 1630
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.3800494641385e-05,
      "loss": 1.3377,
      "step": 1640
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.4006595218466616e-05,
      "loss": 1.2925,
      "step": 1650
    },
    {
      "epoch": 0.21,
      "learning_rate": 3.421269579554823e-05,
      "loss": 1.3274,
      "step": 1660
    },
    {
      "epoch": 0.21,
      "learning_rate": 3.441879637262985e-05,
      "loss": 1.2687,
      "step": 1670
    },
    {
      "epoch": 0.21,
      "learning_rate": 3.4624896949711464e-05,
      "loss": 1.3106,
      "step": 1680
    },
    {
      "epoch": 0.21,
      "learning_rate": 3.483099752679308e-05,
      "loss": 1.3474,
      "step": 1690
    },
    {
      "epoch": 0.21,
      "learning_rate": 3.503709810387469e-05,
      "loss": 1.2755,
      "step": 1700
    },
    {
      "epoch": 0.21,
      "learning_rate": 3.5243198680956305e-05,
      "loss": 1.3411,
      "step": 1710
    },
    {
      "epoch": 0.21,
      "learning_rate": 3.5449299258037925e-05,
      "loss": 1.3302,
      "step": 1720
    },
    {
      "epoch": 0.21,
      "learning_rate": 3.565539983511954e-05,
      "loss": 1.3389,
      "step": 1730
    },
    {
      "epoch": 0.22,
      "learning_rate": 3.586150041220115e-05,
      "loss": 1.275,
      "step": 1740
    },
    {
      "epoch": 0.22,
      "learning_rate": 3.606760098928277e-05,
      "loss": 1.3162,
      "step": 1750
    },
    {
      "epoch": 0.22,
      "learning_rate": 3.6273701566364386e-05,
      "loss": 1.3057,
      "step": 1760
    },
    {
      "epoch": 0.22,
      "learning_rate": 3.6479802143446e-05,
      "loss": 1.2945,
      "step": 1770
    },
    {
      "epoch": 0.22,
      "learning_rate": 3.668590272052762e-05,
      "loss": 1.2761,
      "step": 1780
    },
    {
      "epoch": 0.22,
      "learning_rate": 3.6892003297609234e-05,
      "loss": 1.2827,
      "step": 1790
    },
    {
      "epoch": 0.22,
      "learning_rate": 3.709810387469085e-05,
      "loss": 1.2988,
      "step": 1800
    },
    {
      "epoch": 0.22,
      "learning_rate": 3.730420445177247e-05,
      "loss": 1.3273,
      "step": 1810
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.751030502885408e-05,
      "loss": 1.2644,
      "step": 1820
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.77164056059357e-05,
      "loss": 1.3081,
      "step": 1830
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.7922506183017316e-05,
      "loss": 1.3072,
      "step": 1840
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.812860676009893e-05,
      "loss": 1.2689,
      "step": 1850
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.833470733718055e-05,
      "loss": 1.2613,
      "step": 1860
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.8540807914262164e-05,
      "loss": 1.3147,
      "step": 1870
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.874690849134378e-05,
      "loss": 1.2822,
      "step": 1880
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.89530090684254e-05,
      "loss": 1.2943,
      "step": 1890
    },
    {
      "epoch": 0.24,
      "learning_rate": 3.915910964550701e-05,
      "loss": 1.2895,
      "step": 1900
    },
    {
      "epoch": 0.24,
      "learning_rate": 3.9365210222588625e-05,
      "loss": 1.2749,
      "step": 1910
    },
    {
      "epoch": 0.24,
      "learning_rate": 3.957131079967024e-05,
      "loss": 1.2827,
      "step": 1920
    },
    {
      "epoch": 0.24,
      "learning_rate": 3.977741137675185e-05,
      "loss": 1.2815,
      "step": 1930
    },
    {
      "epoch": 0.24,
      "learning_rate": 3.998351195383347e-05,
      "loss": 1.2576,
      "step": 1940
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.0189612530915086e-05,
      "loss": 1.2843,
      "step": 1950
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.03957131079967e-05,
      "loss": 1.2842,
      "step": 1960
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.060181368507832e-05,
      "loss": 1.2971,
      "step": 1970
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.0807914262159934e-05,
      "loss": 1.2993,
      "step": 1980
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.101401483924155e-05,
      "loss": 1.297,
      "step": 1990
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.122011541632317e-05,
      "loss": 1.2657,
      "step": 2000
    },
    {
      "epoch": 0.25,
      "eval_loss": 1.1918519735336304,
      "eval_runtime": 5.7665,
      "eval_samples_per_second": 89.829,
      "eval_steps_per_second": 11.272,
      "step": 2000
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.142621599340478e-05,
      "loss": 1.2862,
      "step": 2010
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.16323165704864e-05,
      "loss": 1.2709,
      "step": 2020
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.1838417147568016e-05,
      "loss": 1.2649,
      "step": 2030
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.204451772464963e-05,
      "loss": 1.317,
      "step": 2040
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.225061830173125e-05,
      "loss": 1.2622,
      "step": 2050
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.245671887881286e-05,
      "loss": 1.2988,
      "step": 2060
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.266281945589448e-05,
      "loss": 1.281,
      "step": 2070
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.28689200329761e-05,
      "loss": 1.2865,
      "step": 2080
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.307502061005771e-05,
      "loss": 1.2695,
      "step": 2090
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.328112118713933e-05,
      "loss": 1.2806,
      "step": 2100
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.3487221764220945e-05,
      "loss": 1.291,
      "step": 2110
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.369332234130256e-05,
      "loss": 1.2854,
      "step": 2120
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.389942291838417e-05,
      "loss": 1.2617,
      "step": 2130
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.4105523495465786e-05,
      "loss": 1.3352,
      "step": 2140
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.4311624072547406e-05,
      "loss": 1.2832,
      "step": 2150
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.451772464962902e-05,
      "loss": 1.2778,
      "step": 2160
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.4723825226710634e-05,
      "loss": 1.2679,
      "step": 2170
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.4929925803792254e-05,
      "loss": 1.294,
      "step": 2180
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.513602638087387e-05,
      "loss": 1.265,
      "step": 2190
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.534212695795548e-05,
      "loss": 1.2549,
      "step": 2200
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.55482275350371e-05,
      "loss": 1.2885,
      "step": 2210
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.5754328112118715e-05,
      "loss": 1.2858,
      "step": 2220
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.596042868920033e-05,
      "loss": 1.2742,
      "step": 2230
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.616652926628195e-05,
      "loss": 1.3061,
      "step": 2240
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.637262984336356e-05,
      "loss": 1.3293,
      "step": 2250
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.657873042044518e-05,
      "loss": 1.2671,
      "step": 2260
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.67848309975268e-05,
      "loss": 1.2881,
      "step": 2270
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.699093157460841e-05,
      "loss": 1.2959,
      "step": 2280
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.719703215169003e-05,
      "loss": 1.3,
      "step": 2290
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.7403132728771645e-05,
      "loss": 1.2194,
      "step": 2300
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.760923330585326e-05,
      "loss": 1.2769,
      "step": 2310
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.781533388293488e-05,
      "loss": 1.2252,
      "step": 2320
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.802143446001649e-05,
      "loss": 1.2456,
      "step": 2330
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.8227535037098106e-05,
      "loss": 1.3035,
      "step": 2340
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.8433635614179727e-05,
      "loss": 1.2018,
      "step": 2350
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.863973619126134e-05,
      "loss": 1.2896,
      "step": 2360
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.8845836768342954e-05,
      "loss": 1.2485,
      "step": 2370
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.905193734542457e-05,
      "loss": 1.2543,
      "step": 2380
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.925803792250618e-05,
      "loss": 1.2763,
      "step": 2390
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.94641384995878e-05,
      "loss": 1.2606,
      "step": 2400
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.9670239076669415e-05,
      "loss": 1.2606,
      "step": 2410
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.987633965375103e-05,
      "loss": 1.2778,
      "step": 2420
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.9999995856365626e-05,
      "loss": 1.2752,
      "step": 2430
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.999994924049467e-05,
      "loss": 1.2708,
      "step": 2440
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.999985082930671e-05,
      "loss": 1.257,
      "step": 2450
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.999970062300561e-05,
      "loss": 1.2547,
      "step": 2460
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.999949862190258e-05,
      "loss": 1.26,
      "step": 2470
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.999924482641613e-05,
      "loss": 1.2638,
      "step": 2480
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.999893923707209e-05,
      "loss": 1.2996,
      "step": 2490
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.9998581854503555e-05,
      "loss": 1.2208,
      "step": 2500
    },
    {
      "epoch": 0.31,
      "eval_loss": 1.1681606769561768,
      "eval_runtime": 5.7701,
      "eval_samples_per_second": 89.774,
      "eval_steps_per_second": 11.265,
      "step": 2500
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.9998172679450984e-05,
      "loss": 1.2741,
      "step": 2510
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.9997711712762105e-05,
      "loss": 1.2765,
      "step": 2520
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.9997198955391945e-05,
      "loss": 1.2402,
      "step": 2530
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.9996634408402866e-05,
      "loss": 1.2759,
      "step": 2540
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.9996018072964485e-05,
      "loss": 1.2466,
      "step": 2550
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.999534995035374e-05,
      "loss": 1.3292,
      "step": 2560
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.999463004195487e-05,
      "loss": 1.2591,
      "step": 2570
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.999385834925939e-05,
      "loss": 1.2557,
      "step": 2580
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.99930348738661e-05,
      "loss": 1.2687,
      "step": 2590
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.999215961748109e-05,
      "loss": 1.2432,
      "step": 2600
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.999123258191774e-05,
      "loss": 1.2748,
      "step": 2610
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.999025376909668e-05,
      "loss": 1.2481,
      "step": 2620
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.998922318104585e-05,
      "loss": 1.2332,
      "step": 2630
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.998814081990043e-05,
      "loss": 1.201,
      "step": 2640
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.998700668790287e-05,
      "loss": 1.318,
      "step": 2650
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.9985820787402905e-05,
      "loss": 1.2837,
      "step": 2660
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.9984583120857476e-05,
      "loss": 1.2287,
      "step": 2670
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.9983293690830816e-05,
      "loss": 1.2462,
      "step": 2680
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.9981952499994386e-05,
      "loss": 1.2682,
      "step": 2690
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.9980559551126895e-05,
      "loss": 1.2765,
      "step": 2700
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.9979114847114265e-05,
      "loss": 1.2422,
      "step": 2710
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.9977618390949666e-05,
      "loss": 1.2488,
      "step": 2720
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.9976070185733484e-05,
      "loss": 1.2046,
      "step": 2730
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.9974470234673296e-05,
      "loss": 1.3038,
      "step": 2740
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.9972818541083935e-05,
      "loss": 1.268,
      "step": 2750
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.99711151083874e-05,
      "loss": 1.2072,
      "step": 2760
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.996935994011288e-05,
      "loss": 1.2273,
      "step": 2770
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.9967553039896774e-05,
      "loss": 1.2566,
      "step": 2780
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.996569441148264e-05,
      "loss": 1.2605,
      "step": 2790
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.996378405872123e-05,
      "loss": 1.2316,
      "step": 2800
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.996182198557043e-05,
      "loss": 1.264,
      "step": 2810
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.99598081960953e-05,
      "loss": 1.2703,
      "step": 2820
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.995774269446805e-05,
      "loss": 1.2591,
      "step": 2830
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.9955625484968004e-05,
      "loss": 1.2575,
      "step": 2840
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.995345657198165e-05,
      "loss": 1.2471,
      "step": 2850
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.995123596000258e-05,
      "loss": 1.1949,
      "step": 2860
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.9948963653631484e-05,
      "loss": 1.2396,
      "step": 2870
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.994663965757617e-05,
      "loss": 1.2551,
      "step": 2880
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.994426397665152e-05,
      "loss": 1.2288,
      "step": 2890
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.994183661577954e-05,
      "loss": 1.2325,
      "step": 2900
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.993935757998924e-05,
      "loss": 1.2827,
      "step": 2910
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.9936826874416754e-05,
      "loss": 1.2477,
      "step": 2920
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.993424450430524e-05,
      "loss": 1.2608,
      "step": 2930
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.993161047500488e-05,
      "loss": 1.2493,
      "step": 2940
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.992892479197291e-05,
      "loss": 1.2592,
      "step": 2950
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.992618746077358e-05,
      "loss": 1.2694,
      "step": 2960
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.992339848707812e-05,
      "loss": 1.2459,
      "step": 2970
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.9920557876664796e-05,
      "loss": 1.2578,
      "step": 2980
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.991766563541882e-05,
      "loss": 1.2377,
      "step": 2990
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.991472176933238e-05,
      "loss": 1.2454,
      "step": 3000
    },
    {
      "epoch": 0.37,
      "eval_loss": 1.1485267877578735,
      "eval_runtime": 5.7624,
      "eval_samples_per_second": 89.893,
      "eval_steps_per_second": 11.28,
      "step": 3000
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.991172628450464e-05,
      "loss": 1.2127,
      "step": 3010
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.990867918714169e-05,
      "loss": 1.2452,
      "step": 3020
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.990558048355656e-05,
      "loss": 1.2212,
      "step": 3030
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.9902430180169194e-05,
      "loss": 1.2321,
      "step": 3040
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.989922828350645e-05,
      "loss": 1.2246,
      "step": 3050
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.989597480020206e-05,
      "loss": 1.2435,
      "step": 3060
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.989266973699666e-05,
      "loss": 1.25,
      "step": 3070
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.9889313100737716e-05,
      "loss": 1.2689,
      "step": 3080
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.9885904898379584e-05,
      "loss": 1.2067,
      "step": 3090
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.988244513698343e-05,
      "loss": 1.2217,
      "step": 3100
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.987893382371723e-05,
      "loss": 1.2208,
      "step": 3110
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.9875370965855805e-05,
      "loss": 1.2536,
      "step": 3120
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.987175657078074e-05,
      "loss": 1.2553,
      "step": 3130
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.9868090645980384e-05,
      "loss": 1.2373,
      "step": 3140
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.986437319904987e-05,
      "loss": 1.2663,
      "step": 3150
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.986060423769108e-05,
      "loss": 1.1748,
      "step": 3160
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.985678376971259e-05,
      "loss": 1.2302,
      "step": 3170
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.985291180302972e-05,
      "loss": 1.2281,
      "step": 3180
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.984898834566449e-05,
      "loss": 1.2421,
      "step": 3190
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.984501340574556e-05,
      "loss": 1.2311,
      "step": 3200
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.9840986991508295e-05,
      "loss": 1.2272,
      "step": 3210
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.983690911129468e-05,
      "loss": 1.2348,
      "step": 3220
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.983277977355335e-05,
      "loss": 1.2439,
      "step": 3230
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.982859898683952e-05,
      "loss": 1.1972,
      "step": 3240
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.982436675981501e-05,
      "loss": 1.2263,
      "step": 3250
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.982008310124824e-05,
      "loss": 1.2558,
      "step": 3260
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.981574802001415e-05,
      "loss": 1.217,
      "step": 3270
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.9811361525094246e-05,
      "loss": 1.2253,
      "step": 3280
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.980692362557654e-05,
      "loss": 1.2135,
      "step": 3290
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.9802434330655536e-05,
      "loss": 1.227,
      "step": 3300
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.979789364963224e-05,
      "loss": 1.2456,
      "step": 3310
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.979330159191411e-05,
      "loss": 1.2393,
      "step": 3320
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.978865816701505e-05,
      "loss": 1.2548,
      "step": 3330
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.9783963384555386e-05,
      "loss": 1.2317,
      "step": 3340
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.977921725426185e-05,
      "loss": 1.2561,
      "step": 3350
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.977441978596754e-05,
      "loss": 1.2392,
      "step": 3360
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.976957098961195e-05,
      "loss": 1.1877,
      "step": 3370
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.976467087524089e-05,
      "loss": 1.1954,
      "step": 3380
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.975971945300651e-05,
      "loss": 1.2321,
      "step": 3390
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.975471673316724e-05,
      "loss": 1.2348,
      "step": 3400
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.9749662726087795e-05,
      "loss": 1.2135,
      "step": 3410
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.974455744223916e-05,
      "loss": 1.2201,
      "step": 3420
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.973940089219854e-05,
      "loss": 1.2252,
      "step": 3430
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.973419308664938e-05,
      "loss": 1.221,
      "step": 3440
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.972893403638128e-05,
      "loss": 1.2328,
      "step": 3450
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.9723623752290044e-05,
      "loss": 1.213,
      "step": 3460
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.97182622453776e-05,
      "loss": 1.2168,
      "step": 3470
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.9712849526752006e-05,
      "loss": 1.2423,
      "step": 3480
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.970738560762743e-05,
      "loss": 1.2456,
      "step": 3490
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.9701870499324124e-05,
      "loss": 1.2492,
      "step": 3500
    },
    {
      "epoch": 0.43,
      "eval_loss": 1.128912329673767,
      "eval_runtime": 5.7611,
      "eval_samples_per_second": 89.914,
      "eval_steps_per_second": 11.283,
      "step": 3500
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.969630421326835e-05,
      "loss": 1.2526,
      "step": 3510
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.969068676099246e-05,
      "loss": 1.2282,
      "step": 3520
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.9685018154134774e-05,
      "loss": 1.2369,
      "step": 3530
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.967929840443962e-05,
      "loss": 1.2392,
      "step": 3540
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.9673527523757265e-05,
      "loss": 1.2325,
      "step": 3550
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.966770552404391e-05,
      "loss": 1.2265,
      "step": 3560
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.9661832417361675e-05,
      "loss": 1.1968,
      "step": 3570
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.965590821587858e-05,
      "loss": 1.2021,
      "step": 3580
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.9649932931868455e-05,
      "loss": 1.215,
      "step": 3590
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.964390657771102e-05,
      "loss": 1.2036,
      "step": 3600
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.9637829165891755e-05,
      "loss": 1.2095,
      "step": 3610
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.963170070900197e-05,
      "loss": 1.179,
      "step": 3620
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.962552121973869e-05,
      "loss": 1.2714,
      "step": 3630
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.9619290710904684e-05,
      "loss": 1.1883,
      "step": 3640
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.9613009195408426e-05,
      "loss": 1.2114,
      "step": 3650
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.9606676686264076e-05,
      "loss": 1.2138,
      "step": 3660
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.960029319659142e-05,
      "loss": 1.2177,
      "step": 3670
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.9593858739615886e-05,
      "loss": 1.2479,
      "step": 3680
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.9587373328668494e-05,
      "loss": 1.1912,
      "step": 3690
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.958083697718583e-05,
      "loss": 1.237,
      "step": 3700
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.957424969871e-05,
      "loss": 1.2338,
      "step": 3710
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.956761150688866e-05,
      "loss": 1.2546,
      "step": 3720
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.956092241547491e-05,
      "loss": 1.2343,
      "step": 3730
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.955418243832734e-05,
      "loss": 1.2286,
      "step": 3740
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.954739158940993e-05,
      "loss": 1.2078,
      "step": 3750
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.9540549882792095e-05,
      "loss": 1.1959,
      "step": 3760
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.953365733264859e-05,
      "loss": 1.2023,
      "step": 3770
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.95267139532595e-05,
      "loss": 1.2087,
      "step": 3780
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.9519719759010264e-05,
      "loss": 1.2019,
      "step": 3790
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.9512674764391554e-05,
      "loss": 1.2305,
      "step": 3800
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.9505578983999316e-05,
      "loss": 1.2218,
      "step": 3810
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.94984324325347e-05,
      "loss": 1.1838,
      "step": 3820
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.949123512480406e-05,
      "loss": 1.2167,
      "step": 3830
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.94839870757189e-05,
      "loss": 1.2035,
      "step": 3840
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.947668830029584e-05,
      "loss": 1.1523,
      "step": 3850
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.946933881365662e-05,
      "loss": 1.2367,
      "step": 3860
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.946193863102802e-05,
      "loss": 1.2044,
      "step": 3870
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.9454487767741864e-05,
      "loss": 1.1548,
      "step": 3880
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.944698623923498e-05,
      "loss": 1.2412,
      "step": 3890
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.9439434061049174e-05,
      "loss": 1.2285,
      "step": 3900
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.943183124883115e-05,
      "loss": 1.2423,
      "step": 3910
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.942417781833256e-05,
      "loss": 1.1856,
      "step": 3920
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.9416473785409914e-05,
      "loss": 1.2055,
      "step": 3930
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.940871916602454e-05,
      "loss": 1.2728,
      "step": 3940
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.9400913976242604e-05,
      "loss": 1.1906,
      "step": 3950
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.9393058232235026e-05,
      "loss": 1.2178,
      "step": 3960
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.938515195027746e-05,
      "loss": 1.2145,
      "step": 3970
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.937719514675029e-05,
      "loss": 1.1979,
      "step": 3980
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.936918783813855e-05,
      "loss": 1.1626,
      "step": 3990
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.936113004103192e-05,
      "loss": 1.2156,
      "step": 4000
    },
    {
      "epoch": 0.49,
      "eval_loss": 1.1184728145599365,
      "eval_runtime": 5.7588,
      "eval_samples_per_second": 89.949,
      "eval_steps_per_second": 11.287,
      "step": 4000
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.935302177212468e-05,
      "loss": 1.1915,
      "step": 4010
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.9344863048215665e-05,
      "loss": 1.2098,
      "step": 4020
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.9336653886208265e-05,
      "loss": 1.2082,
      "step": 4030
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.9328394303110365e-05,
      "loss": 1.2193,
      "step": 4040
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.932008431603431e-05,
      "loss": 1.2345,
      "step": 4050
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.931172394219686e-05,
      "loss": 1.1834,
      "step": 4060
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.93033131989192e-05,
      "loss": 1.2171,
      "step": 4070
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.929485210362682e-05,
      "loss": 1.2302,
      "step": 4080
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.9286340673849586e-05,
      "loss": 1.186,
      "step": 4090
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.9277778927221615e-05,
      "loss": 1.1853,
      "step": 4100
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.926916688148127e-05,
      "loss": 1.2002,
      "step": 4110
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.926050455447114e-05,
      "loss": 1.1941,
      "step": 4120
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.925179196413797e-05,
      "loss": 1.2144,
      "step": 4130
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.924302912853268e-05,
      "loss": 1.2232,
      "step": 4140
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.9234216065810226e-05,
      "loss": 1.1838,
      "step": 4150
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.922535279422968e-05,
      "loss": 1.195,
      "step": 4160
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.9216439332154116e-05,
      "loss": 1.184,
      "step": 4170
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.920747569805059e-05,
      "loss": 1.2194,
      "step": 4180
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.9198461910490115e-05,
      "loss": 1.1878,
      "step": 4190
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.91893979881476e-05,
      "loss": 1.1877,
      "step": 4200
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.918028394980184e-05,
      "loss": 1.2224,
      "step": 4210
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.9171119814335456e-05,
      "loss": 1.187,
      "step": 4220
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.9161905600734856e-05,
      "loss": 1.2038,
      "step": 4230
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.9152641328090196e-05,
      "loss": 1.2215,
      "step": 4240
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.914332701559536e-05,
      "loss": 1.234,
      "step": 4250
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.91339626825479e-05,
      "loss": 1.2328,
      "step": 4260
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.9124548348349e-05,
      "loss": 1.1991,
      "step": 4270
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.911508403250342e-05,
      "loss": 1.2052,
      "step": 4280
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.910556975461952e-05,
      "loss": 1.1817,
      "step": 4290
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.9096005534409105e-05,
      "loss": 1.2053,
      "step": 4300
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.908639139168752e-05,
      "loss": 1.2306,
      "step": 4310
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.907672734637349e-05,
      "loss": 1.2283,
      "step": 4320
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.906701341848915e-05,
      "loss": 1.1788,
      "step": 4330
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.905724962815998e-05,
      "loss": 1.1947,
      "step": 4340
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.9047435995614774e-05,
      "loss": 1.1843,
      "step": 4350
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.9037572541185575e-05,
      "loss": 1.2393,
      "step": 4360
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.9027659285307645e-05,
      "loss": 1.2127,
      "step": 4370
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.901769624851946e-05,
      "loss": 1.1986,
      "step": 4380
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.900768345146258e-05,
      "loss": 1.1879,
      "step": 4390
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.8997620914881704e-05,
      "loss": 1.2236,
      "step": 4400
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.8987508659624556e-05,
      "loss": 1.2315,
      "step": 4410
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.8977346706641894e-05,
      "loss": 1.1853,
      "step": 4420
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.8967135076987404e-05,
      "loss": 1.1835,
      "step": 4430
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.8956873791817714e-05,
      "loss": 1.1746,
      "step": 4440
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.894656287239234e-05,
      "loss": 1.197,
      "step": 4450
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.893620234007361e-05,
      "loss": 1.1854,
      "step": 4460
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.892579221632666e-05,
      "loss": 1.2002,
      "step": 4470
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.891533252271935e-05,
      "loss": 1.1718,
      "step": 4480
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.8904823280922254e-05,
      "loss": 1.1758,
      "step": 4490
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.88942645127086e-05,
      "loss": 1.1966,
      "step": 4500
    },
    {
      "epoch": 0.56,
      "eval_loss": 1.1033425331115723,
      "eval_runtime": 5.7651,
      "eval_samples_per_second": 89.852,
      "eval_steps_per_second": 11.275,
      "step": 4500
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.888365623995421e-05,
      "loss": 1.2163,
      "step": 4510
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.8872998484637496e-05,
      "loss": 1.1739,
      "step": 4520
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.8862291268839364e-05,
      "loss": 1.1955,
      "step": 4530
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.885153461474322e-05,
      "loss": 1.199,
      "step": 4540
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.8840728544634865e-05,
      "loss": 1.1892,
      "step": 4550
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.882987308090251e-05,
      "loss": 1.2132,
      "step": 4560
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.881896824603668e-05,
      "loss": 1.1562,
      "step": 4570
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.8808014062630195e-05,
      "loss": 1.204,
      "step": 4580
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.8797010553378125e-05,
      "loss": 1.1742,
      "step": 4590
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.8785957741077724e-05,
      "loss": 1.1683,
      "step": 4600
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.877485564862839e-05,
      "loss": 1.2463,
      "step": 4610
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.876370429903163e-05,
      "loss": 1.1703,
      "step": 4620
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.875250371539101e-05,
      "loss": 1.1976,
      "step": 4630
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.8741253920912054e-05,
      "loss": 1.2428,
      "step": 4640
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.8729954938902314e-05,
      "loss": 1.1834,
      "step": 4650
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.871860679277119e-05,
      "loss": 1.2081,
      "step": 4660
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.870720950602997e-05,
      "loss": 1.2135,
      "step": 4670
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.8695763102291745e-05,
      "loss": 1.1742,
      "step": 4680
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.868426760527136e-05,
      "loss": 1.1693,
      "step": 4690
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.86727230387854e-05,
      "loss": 1.1987,
      "step": 4700
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.866112942675206e-05,
      "loss": 1.1927,
      "step": 4710
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.8649486793191214e-05,
      "loss": 1.223,
      "step": 4720
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.8637795162224255e-05,
      "loss": 1.184,
      "step": 4730
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.8626054558074094e-05,
      "loss": 1.1742,
      "step": 4740
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.861426500506512e-05,
      "loss": 1.1632,
      "step": 4750
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.8602426527623116e-05,
      "loss": 1.2006,
      "step": 4760
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.859053915027526e-05,
      "loss": 1.213,
      "step": 4770
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.857860289765002e-05,
      "loss": 1.2422,
      "step": 4780
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.85666177944771e-05,
      "loss": 1.1855,
      "step": 4790
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.8554583865587466e-05,
      "loss": 1.1988,
      "step": 4800
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.85425011359132e-05,
      "loss": 1.192,
      "step": 4810
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.853036963048752e-05,
      "loss": 1.248,
      "step": 4820
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.851818937444466e-05,
      "loss": 1.1865,
      "step": 4830
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.85059603930199e-05,
      "loss": 1.2067,
      "step": 4840
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.849368271154945e-05,
      "loss": 1.1889,
      "step": 4850
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.8481356355470406e-05,
      "loss": 1.1954,
      "step": 4860
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.846898135032073e-05,
      "loss": 1.2122,
      "step": 4870
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.845655772173916e-05,
      "loss": 1.1761,
      "step": 4880
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.844408549546519e-05,
      "loss": 1.2051,
      "step": 4890
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.843156469733898e-05,
      "loss": 1.1774,
      "step": 4900
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.841899535330133e-05,
      "loss": 1.211,
      "step": 4910
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.840637748939363e-05,
      "loss": 1.1574,
      "step": 4920
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.839371113175777e-05,
      "loss": 1.1925,
      "step": 4930
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.838099630663613e-05,
      "loss": 1.2132,
      "step": 4940
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.83682330403715e-05,
      "loss": 1.2053,
      "step": 4950
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.8355421359407035e-05,
      "loss": 1.2115,
      "step": 4960
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.834256129028618e-05,
      "loss": 1.2162,
      "step": 4970
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.8329652859652655e-05,
      "loss": 1.198,
      "step": 4980
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.831669609425035e-05,
      "loss": 1.1876,
      "step": 4990
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.830369102092333e-05,
      "loss": 1.1742,
      "step": 5000
    },
    {
      "epoch": 0.62,
      "eval_loss": 1.0941241979599,
      "eval_runtime": 5.7612,
      "eval_samples_per_second": 89.912,
      "eval_steps_per_second": 11.282,
      "step": 5000
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.829063766661571e-05,
      "loss": 1.1852,
      "step": 5010
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.827753605837166e-05,
      "loss": 1.1893,
      "step": 5020
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.8264386223335314e-05,
      "loss": 1.2032,
      "step": 5030
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.8251188188750705e-05,
      "loss": 1.1534,
      "step": 5040
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.823794198196176e-05,
      "loss": 1.1918,
      "step": 5050
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.822464763041219e-05,
      "loss": 1.1831,
      "step": 5060
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.821130516164546e-05,
      "loss": 1.2074,
      "step": 5070
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.819791460330472e-05,
      "loss": 1.1676,
      "step": 5080
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.818447598313275e-05,
      "loss": 1.1626,
      "step": 5090
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.817098932897191e-05,
      "loss": 1.2068,
      "step": 5100
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.815745466876409e-05,
      "loss": 1.1729,
      "step": 5110
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.814387203055061e-05,
      "loss": 1.207,
      "step": 5120
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.813024144247223e-05,
      "loss": 1.1348,
      "step": 5130
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.8116562932769014e-05,
      "loss": 1.2031,
      "step": 5140
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.8102836529780336e-05,
      "loss": 1.1794,
      "step": 5150
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.808906226194479e-05,
      "loss": 1.1771,
      "step": 5160
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.807524015780015e-05,
      "loss": 1.2,
      "step": 5170
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.806137024598326e-05,
      "loss": 1.1804,
      "step": 5180
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.804745255523005e-05,
      "loss": 1.1825,
      "step": 5190
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.803348711437543e-05,
      "loss": 1.1929,
      "step": 5200
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.801947395235323e-05,
      "loss": 1.1898,
      "step": 5210
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.800541309819616e-05,
      "loss": 1.151,
      "step": 5220
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.799130458103572e-05,
      "loss": 1.1799,
      "step": 5230
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.7977148430102205e-05,
      "loss": 1.1825,
      "step": 5240
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.796294467472454e-05,
      "loss": 1.1978,
      "step": 5250
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.794869334433032e-05,
      "loss": 1.1872,
      "step": 5260
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.793439446844568e-05,
      "loss": 1.196,
      "step": 5270
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.792004807669529e-05,
      "loss": 1.119,
      "step": 5280
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.790565419880223e-05,
      "loss": 1.1771,
      "step": 5290
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.789121286458798e-05,
      "loss": 1.1829,
      "step": 5300
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.787672410397235e-05,
      "loss": 1.1739,
      "step": 5310
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.7862187946973404e-05,
      "loss": 1.2195,
      "step": 5320
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.784760442370737e-05,
      "loss": 1.1763,
      "step": 5330
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.783297356438867e-05,
      "loss": 1.1886,
      "step": 5340
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.781829539932975e-05,
      "loss": 1.2119,
      "step": 5350
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.7803569958941076e-05,
      "loss": 1.1745,
      "step": 5360
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.778879727373107e-05,
      "loss": 1.1848,
      "step": 5370
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.777397737430603e-05,
      "loss": 1.1866,
      "step": 5380
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.775911029137008e-05,
      "loss": 1.1717,
      "step": 5390
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.774419605572508e-05,
      "loss": 1.1806,
      "step": 5400
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.772923469827061e-05,
      "loss": 1.1827,
      "step": 5410
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.7714226250003844e-05,
      "loss": 1.1779,
      "step": 5420
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.769917074201956e-05,
      "loss": 1.1996,
      "step": 5430
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.7684068205510006e-05,
      "loss": 1.2105,
      "step": 5440
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.766891867176487e-05,
      "loss": 1.2059,
      "step": 5450
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.765372217217121e-05,
      "loss": 1.1851,
      "step": 5460
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.7638478738213386e-05,
      "loss": 1.1709,
      "step": 5470
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.762318840147302e-05,
      "loss": 1.1775,
      "step": 5480
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.7607851193628873e-05,
      "loss": 1.1889,
      "step": 5490
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.759246714645684e-05,
      "loss": 1.2105,
      "step": 5500
    },
    {
      "epoch": 0.68,
      "eval_loss": 1.083298921585083,
      "eval_runtime": 5.7617,
      "eval_samples_per_second": 89.904,
      "eval_steps_per_second": 11.281,
      "step": 5500
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.757703629182984e-05,
      "loss": 1.1806,
      "step": 5510
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.7561558661717785e-05,
      "loss": 1.1953,
      "step": 5520
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.75460342881875e-05,
      "loss": 1.1784,
      "step": 5530
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.753046320340262e-05,
      "loss": 1.1747,
      "step": 5540
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.751484543962361e-05,
      "loss": 1.1441,
      "step": 5550
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.7499181029207596e-05,
      "loss": 1.1629,
      "step": 5560
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.748347000460837e-05,
      "loss": 1.1665,
      "step": 5570
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.7467712398376304e-05,
      "loss": 1.1578,
      "step": 5580
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.745190824315828e-05,
      "loss": 1.1769,
      "step": 5590
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.7436057571697604e-05,
      "loss": 1.1652,
      "step": 5600
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.7420160416833954e-05,
      "loss": 1.1147,
      "step": 5610
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.7404216811503344e-05,
      "loss": 1.1666,
      "step": 5620
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.738822678873799e-05,
      "loss": 1.1638,
      "step": 5630
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.73721903816663e-05,
      "loss": 1.1673,
      "step": 5640
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.735610762351278e-05,
      "loss": 1.1946,
      "step": 5650
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.733997854759794e-05,
      "loss": 1.1892,
      "step": 5660
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.732380318733829e-05,
      "loss": 1.1751,
      "step": 5670
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.730758157624619e-05,
      "loss": 1.1635,
      "step": 5680
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.729131374792989e-05,
      "loss": 1.151,
      "step": 5690
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.72749997360933e-05,
      "loss": 1.1779,
      "step": 5700
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.7258639574536114e-05,
      "loss": 1.1868,
      "step": 5710
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.724223329715356e-05,
      "loss": 1.2185,
      "step": 5720
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.7225780937936445e-05,
      "loss": 1.1568,
      "step": 5730
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.720928253097105e-05,
      "loss": 1.175,
      "step": 5740
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.719273811043906e-05,
      "loss": 1.1683,
      "step": 5750
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.717614771061747e-05,
      "loss": 1.1569,
      "step": 5760
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.7159511365878556e-05,
      "loss": 1.1615,
      "step": 5770
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.714282911068978e-05,
      "loss": 1.1535,
      "step": 5780
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.712610097961372e-05,
      "loss": 1.1987,
      "step": 5790
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.7109327007308004e-05,
      "loss": 1.17,
      "step": 5800
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.7092507228525216e-05,
      "loss": 1.1645,
      "step": 5810
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.707564167811287e-05,
      "loss": 1.1478,
      "step": 5820
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.7058730391013305e-05,
      "loss": 1.1658,
      "step": 5830
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.70417734022636e-05,
      "loss": 1.1564,
      "step": 5840
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.7024770746995526e-05,
      "loss": 1.184,
      "step": 5850
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.700772246043549e-05,
      "loss": 1.1794,
      "step": 5860
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.69906285779044e-05,
      "loss": 1.1645,
      "step": 5870
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.6973489134817664e-05,
      "loss": 1.1377,
      "step": 5880
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.695630416668506e-05,
      "loss": 1.1282,
      "step": 5890
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.693907370911071e-05,
      "loss": 1.2042,
      "step": 5900
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.6921797797792944e-05,
      "loss": 1.1495,
      "step": 5910
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.690447646852431e-05,
      "loss": 1.1368,
      "step": 5920
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.6887109757191413e-05,
      "loss": 1.1746,
      "step": 5930
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.686969769977491e-05,
      "loss": 1.1704,
      "step": 5940
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.6852240332349384e-05,
      "loss": 1.1368,
      "step": 5950
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.683473769108333e-05,
      "loss": 1.168,
      "step": 5960
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.6817189812238975e-05,
      "loss": 1.1668,
      "step": 5970
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.6799596732172346e-05,
      "loss": 1.16,
      "step": 5980
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.6781958487333066e-05,
      "loss": 1.1758,
      "step": 5990
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.676427511426435e-05,
      "loss": 1.15,
      "step": 6000
    },
    {
      "epoch": 0.74,
      "eval_loss": 1.0752549171447754,
      "eval_runtime": 5.7664,
      "eval_samples_per_second": 89.831,
      "eval_steps_per_second": 11.272,
      "step": 6000
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.674654664960292e-05,
      "loss": 1.1545,
      "step": 6010
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.6728773130078895e-05,
      "loss": 1.1549,
      "step": 6020
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.671095459251576e-05,
      "loss": 1.1818,
      "step": 6030
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.6693091073830255e-05,
      "loss": 1.1497,
      "step": 6040
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.667518261103233e-05,
      "loss": 1.1583,
      "step": 6050
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.665722924122504e-05,
      "loss": 1.1194,
      "step": 6060
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.663923100160447e-05,
      "loss": 1.1617,
      "step": 6070
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.662118792945967e-05,
      "loss": 1.1584,
      "step": 6080
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.660310006217261e-05,
      "loss": 1.1676,
      "step": 6090
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.6584967437218016e-05,
      "loss": 1.176,
      "step": 6100
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.6566790092163373e-05,
      "loss": 1.1522,
      "step": 6110
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.654856806466882e-05,
      "loss": 1.1381,
      "step": 6120
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.653030139248704e-05,
      "loss": 1.1316,
      "step": 6130
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.651199011346325e-05,
      "loss": 1.1556,
      "step": 6140
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.649363426553506e-05,
      "loss": 1.2094,
      "step": 6150
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.647523388673244e-05,
      "loss": 1.1663,
      "step": 6160
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.645678901517758e-05,
      "loss": 1.1911,
      "step": 6170
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.6438299689084894e-05,
      "loss": 1.1605,
      "step": 6180
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.641976594676088e-05,
      "loss": 1.1717,
      "step": 6190
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.6401187826604054e-05,
      "loss": 1.1539,
      "step": 6200
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.638256536710488e-05,
      "loss": 1.1711,
      "step": 6210
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.636389860684568e-05,
      "loss": 1.174,
      "step": 6220
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.634518758450057e-05,
      "loss": 1.1408,
      "step": 6230
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.632643233883537e-05,
      "loss": 1.1988,
      "step": 6240
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.630763290870749e-05,
      "loss": 1.1776,
      "step": 6250
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.628878933306592e-05,
      "loss": 1.1396,
      "step": 6260
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.62699016509511e-05,
      "loss": 1.1702,
      "step": 6270
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.625096990149486e-05,
      "loss": 1.1576,
      "step": 6280
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.623199412392029e-05,
      "loss": 1.1564,
      "step": 6290
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.621297435754175e-05,
      "loss": 1.1574,
      "step": 6300
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.61939106417647e-05,
      "loss": 1.1785,
      "step": 6310
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.6174803016085676e-05,
      "loss": 1.1694,
      "step": 6320
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.615565152009218e-05,
      "loss": 1.1607,
      "step": 6330
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.6136456193462604e-05,
      "loss": 1.1641,
      "step": 6340
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.611721707596615e-05,
      "loss": 1.1649,
      "step": 6350
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.6097934207462734e-05,
      "loss": 1.1253,
      "step": 6360
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.6078607627902944e-05,
      "loss": 1.1772,
      "step": 6370
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.6059237377327905e-05,
      "loss": 1.146,
      "step": 6380
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.603982349586924e-05,
      "loss": 1.1865,
      "step": 6390
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.6020366023748945e-05,
      "loss": 1.1718,
      "step": 6400
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.6000865001279345e-05,
      "loss": 1.1528,
      "step": 6410
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.598132046886299e-05,
      "loss": 1.1702,
      "step": 6420
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.5961732466992566e-05,
      "loss": 1.1807,
      "step": 6430
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.5942101036250826e-05,
      "loss": 1.1482,
      "step": 6440
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.592242621731051e-05,
      "loss": 1.1516,
      "step": 6450
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.590270805093423e-05,
      "loss": 1.1622,
      "step": 6460
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.588294657797442e-05,
      "loss": 1.1704,
      "step": 6470
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.586314183937324e-05,
      "loss": 1.1556,
      "step": 6480
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.584329387616246e-05,
      "loss": 1.1277,
      "step": 6490
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.582340272946345e-05,
      "loss": 1.104,
      "step": 6500
    },
    {
      "epoch": 0.8,
      "eval_loss": 1.0666757822036743,
      "eval_runtime": 5.7634,
      "eval_samples_per_second": 89.878,
      "eval_steps_per_second": 11.278,
      "step": 6500
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.5803468440487004e-05,
      "loss": 1.153,
      "step": 6510
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.578349105053334e-05,
      "loss": 1.1413,
      "step": 6520
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.576347060099192e-05,
      "loss": 1.147,
      "step": 6530
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.5743407133341484e-05,
      "loss": 1.1875,
      "step": 6540
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.5723300689149844e-05,
      "loss": 1.1913,
      "step": 6550
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.570315131007388e-05,
      "loss": 1.1226,
      "step": 6560
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.568295903785941e-05,
      "loss": 1.1865,
      "step": 6570
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.566272391434112e-05,
      "loss": 1.1709,
      "step": 6580
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.564244598144249e-05,
      "loss": 1.1522,
      "step": 6590
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.562212528117568e-05,
      "loss": 1.1143,
      "step": 6600
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.560176185564146e-05,
      "loss": 1.148,
      "step": 6610
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.558135574702911e-05,
      "loss": 1.1695,
      "step": 6620
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.5560906997616367e-05,
      "loss": 1.1309,
      "step": 6630
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.5540415649769284e-05,
      "loss": 1.1146,
      "step": 6640
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.5519881745942186e-05,
      "loss": 1.1798,
      "step": 6650
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.549930532867757e-05,
      "loss": 1.1346,
      "step": 6660
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.5478686440605984e-05,
      "loss": 1.1633,
      "step": 6670
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.5458025124446005e-05,
      "loss": 1.1527,
      "step": 6680
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.543732142300408e-05,
      "loss": 1.1634,
      "step": 6690
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.5416575379174515e-05,
      "loss": 1.1537,
      "step": 6700
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.539578703593929e-05,
      "loss": 1.1409,
      "step": 6710
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.5374956436368055e-05,
      "loss": 1.1525,
      "step": 6720
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.5354083623618e-05,
      "loss": 1.1341,
      "step": 6730
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.533316864093376e-05,
      "loss": 1.1455,
      "step": 6740
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.5312211531647364e-05,
      "loss": 1.1527,
      "step": 6750
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.529121233917809e-05,
      "loss": 1.1348,
      "step": 6760
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.527017110703243e-05,
      "loss": 1.1579,
      "step": 6770
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.5249087878803965e-05,
      "loss": 1.1542,
      "step": 6780
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.5227962698173274e-05,
      "loss": 1.1334,
      "step": 6790
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.520679560890787e-05,
      "loss": 1.1488,
      "step": 6800
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.518558665486209e-05,
      "loss": 1.1579,
      "step": 6810
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.516433587997699e-05,
      "loss": 1.1086,
      "step": 6820
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.51430433282803e-05,
      "loss": 1.1344,
      "step": 6830
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.512170904388627e-05,
      "loss": 1.1374,
      "step": 6840
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.510033307099565e-05,
      "loss": 1.1784,
      "step": 6850
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.5078915453895535e-05,
      "loss": 1.1594,
      "step": 6860
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.5057456236959296e-05,
      "loss": 1.1497,
      "step": 6870
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.503595546464651e-05,
      "loss": 1.1802,
      "step": 6880
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.501441318150283e-05,
      "loss": 1.1592,
      "step": 6890
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.499282943215994e-05,
      "loss": 1.1789,
      "step": 6900
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.497120426133541e-05,
      "loss": 1.1523,
      "step": 6910
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.494953771383262e-05,
      "loss": 1.2015,
      "step": 6920
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.4927829834540715e-05,
      "loss": 1.1259,
      "step": 6930
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.4906080668434416e-05,
      "loss": 1.1982,
      "step": 6940
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.488429026057404e-05,
      "loss": 1.1435,
      "step": 6950
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.48624586561053e-05,
      "loss": 1.1524,
      "step": 6960
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.4840585900259296e-05,
      "loss": 1.1662,
      "step": 6970
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.481867203835237e-05,
      "loss": 1.1123,
      "step": 6980
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.479671711578603e-05,
      "loss": 1.1497,
      "step": 6990
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.4774721178046855e-05,
      "loss": 1.1224,
      "step": 7000
    },
    {
      "epoch": 0.87,
      "eval_loss": 1.0597686767578125,
      "eval_runtime": 5.764,
      "eval_samples_per_second": 89.867,
      "eval_steps_per_second": 11.277,
      "step": 7000
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.47526842707064e-05,
      "loss": 1.1454,
      "step": 7010
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.47306064394211e-05,
      "loss": 1.1773,
      "step": 7020
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.470848772993219e-05,
      "loss": 1.1513,
      "step": 7030
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.468632818806557e-05,
      "loss": 1.1719,
      "step": 7040
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.4664127859731766e-05,
      "loss": 1.1373,
      "step": 7050
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.464188679092578e-05,
      "loss": 1.1578,
      "step": 7060
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.461960502772704e-05,
      "loss": 1.1745,
      "step": 7070
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.4597282616299294e-05,
      "loss": 1.1643,
      "step": 7080
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.457491960289046e-05,
      "loss": 1.1308,
      "step": 7090
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.4552516033832645e-05,
      "loss": 1.126,
      "step": 7100
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.453007195554191e-05,
      "loss": 1.1931,
      "step": 7110
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.450758741451828e-05,
      "loss": 1.1437,
      "step": 7120
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.448506245734563e-05,
      "loss": 1.1313,
      "step": 7130
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.4462497130691514e-05,
      "loss": 1.1492,
      "step": 7140
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.4439891481307184e-05,
      "loss": 1.1775,
      "step": 7150
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.441724555602739e-05,
      "loss": 1.2022,
      "step": 7160
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.4394559401770344e-05,
      "loss": 1.1165,
      "step": 7170
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.4371833065537614e-05,
      "loss": 1.1326,
      "step": 7180
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.4349066594413996e-05,
      "loss": 1.1371,
      "step": 7190
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.4326260035567456e-05,
      "loss": 1.1254,
      "step": 7200
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.430341343624901e-05,
      "loss": 1.1212,
      "step": 7210
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.4280526843792614e-05,
      "loss": 1.1501,
      "step": 7220
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.425760030561511e-05,
      "loss": 1.0971,
      "step": 7230
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.423463386921608e-05,
      "loss": 1.1613,
      "step": 7240
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.421162758217778e-05,
      "loss": 1.142,
      "step": 7250
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.4188581492165025e-05,
      "loss": 1.1541,
      "step": 7260
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.4165495646925095e-05,
      "loss": 1.1291,
      "step": 7270
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.414237009428762e-05,
      "loss": 1.1866,
      "step": 7280
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.411920488216452e-05,
      "loss": 1.1467,
      "step": 7290
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.409600005854988e-05,
      "loss": 1.1166,
      "step": 7300
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.407275567151983e-05,
      "loss": 1.1589,
      "step": 7310
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.4049471769232496e-05,
      "loss": 1.2047,
      "step": 7320
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.402614839992786e-05,
      "loss": 1.2128,
      "step": 7330
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.400278561192767e-05,
      "loss": 1.1628,
      "step": 7340
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.3979383453635333e-05,
      "loss": 1.1453,
      "step": 7350
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.395594197353585e-05,
      "loss": 1.1652,
      "step": 7360
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.393246122019567e-05,
      "loss": 1.1544,
      "step": 7370
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.390894124226261e-05,
      "loss": 1.0748,
      "step": 7380
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.388538208846577e-05,
      "loss": 1.1389,
      "step": 7390
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.3861783807615396e-05,
      "loss": 1.1769,
      "step": 7400
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.38381464486028e-05,
      "loss": 1.1197,
      "step": 7410
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.381447006040027e-05,
      "loss": 1.1511,
      "step": 7420
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.379075469206094e-05,
      "loss": 1.1547,
      "step": 7430
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.376700039271871e-05,
      "loss": 1.1686,
      "step": 7440
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.374320721158814e-05,
      "loss": 1.1379,
      "step": 7450
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.371937519796436e-05,
      "loss": 1.1357,
      "step": 7460
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.369550440122291e-05,
      "loss": 1.1332,
      "step": 7470
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.3671594870819734e-05,
      "loss": 1.0943,
      "step": 7480
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.3647646656290985e-05,
      "loss": 1.1537,
      "step": 7490
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.362365980725299e-05,
      "loss": 1.164,
      "step": 7500
    },
    {
      "epoch": 0.93,
      "eval_loss": 1.0545625686645508,
      "eval_runtime": 5.7642,
      "eval_samples_per_second": 89.865,
      "eval_steps_per_second": 11.277,
      "step": 7500
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.3599634373402096e-05,
      "loss": 1.17,
      "step": 7510
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.357557040451461e-05,
      "loss": 1.1273,
      "step": 7520
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.3551467950446666e-05,
      "loss": 1.1338,
      "step": 7530
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.352732706113415e-05,
      "loss": 1.1678,
      "step": 7540
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.3503147786592534e-05,
      "loss": 1.1786,
      "step": 7550
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.3478930176916875e-05,
      "loss": 1.1294,
      "step": 7560
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.3454674282281615e-05,
      "loss": 1.1333,
      "step": 7570
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.343038015294053e-05,
      "loss": 1.1515,
      "step": 7580
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.3406047839226596e-05,
      "loss": 1.1936,
      "step": 7590
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.3381677391551926e-05,
      "loss": 1.1203,
      "step": 7600
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.3357268860407616e-05,
      "loss": 1.1492,
      "step": 7610
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.3332822296363675e-05,
      "loss": 1.1298,
      "step": 7620
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.3308337750068906e-05,
      "loss": 1.1431,
      "step": 7630
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.328381527225082e-05,
      "loss": 1.1197,
      "step": 7640
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.3259254913715476e-05,
      "loss": 1.1265,
      "step": 7650
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.323465672534745e-05,
      "loss": 1.1421,
      "step": 7660
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.321002075810969e-05,
      "loss": 1.1673,
      "step": 7670
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.3185347063043405e-05,
      "loss": 1.1151,
      "step": 7680
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.316063569126797e-05,
      "loss": 1.1386,
      "step": 7690
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.3135886693980826e-05,
      "loss": 1.1347,
      "step": 7700
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.3111100122457364e-05,
      "loss": 1.1438,
      "step": 7710
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.308627602805082e-05,
      "loss": 1.1472,
      "step": 7720
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.306141446219217e-05,
      "loss": 1.1036,
      "step": 7730
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.3036515476390033e-05,
      "loss": 1.1359,
      "step": 7740
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.3011579122230536e-05,
      "loss": 1.1645,
      "step": 7750
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.298660545137725e-05,
      "loss": 1.1119,
      "step": 7760
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.296159451557105e-05,
      "loss": 1.1248,
      "step": 7770
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.293654636663e-05,
      "loss": 1.123,
      "step": 7780
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.291146105644929e-05,
      "loss": 1.1446,
      "step": 7790
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.2886338637001086e-05,
      "loss": 1.1314,
      "step": 7800
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.2861179160334445e-05,
      "loss": 1.1581,
      "step": 7810
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.283598267857519e-05,
      "loss": 1.1603,
      "step": 7820
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.281074924392582e-05,
      "loss": 1.1182,
      "step": 7830
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.278547890866538e-05,
      "loss": 1.1635,
      "step": 7840
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.2760171725149404e-05,
      "loss": 1.156,
      "step": 7850
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.2734827745809715e-05,
      "loss": 1.1459,
      "step": 7860
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.2709447023154415e-05,
      "loss": 1.1402,
      "step": 7870
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.26840296097677e-05,
      "loss": 1.1653,
      "step": 7880
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.265857555830981e-05,
      "loss": 1.1699,
      "step": 7890
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.2633084921516864e-05,
      "loss": 1.1403,
      "step": 7900
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.260755775220081e-05,
      "loss": 1.1759,
      "step": 7910
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.2581994103249254e-05,
      "loss": 1.1511,
      "step": 7920
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.2556394027625405e-05,
      "loss": 1.146,
      "step": 7930
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.253075757836793e-05,
      "loss": 1.1145,
      "step": 7940
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.2505084808590855e-05,
      "loss": 1.1548,
      "step": 7950
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.247937577148345e-05,
      "loss": 1.1272,
      "step": 7960
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.245363052031016e-05,
      "loss": 1.1469,
      "step": 7970
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.24278491084104e-05,
      "loss": 1.1679,
      "step": 7980
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.2402031589198554e-05,
      "loss": 1.1287,
      "step": 7990
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.237617801616378e-05,
      "loss": 1.1448,
      "step": 8000
    },
    {
      "epoch": 0.99,
      "eval_loss": 1.0445607900619507,
      "eval_runtime": 5.7627,
      "eval_samples_per_second": 89.889,
      "eval_steps_per_second": 11.279,
      "step": 8000
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.235028844286997e-05,
      "loss": 1.1526,
      "step": 8010
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.2324362922955546e-05,
      "loss": 1.145,
      "step": 8020
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.229840151013347e-05,
      "loss": 1.1585,
      "step": 8030
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.2272404258191014e-05,
      "loss": 1.1796,
      "step": 8040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.224637122098972e-05,
      "loss": 1.148,
      "step": 8050
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.22203024524653e-05,
      "loss": 1.1426,
      "step": 8060
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.219419800662745e-05,
      "loss": 1.1373,
      "step": 8070
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.216805793755981e-05,
      "loss": 1.1189,
      "step": 8080
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.2141882299419806e-05,
      "loss": 1.097,
      "step": 8090
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.2115671146438574e-05,
      "loss": 1.1113,
      "step": 8100
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.208942453292083e-05,
      "loss": 1.1328,
      "step": 8110
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.2063142513244736e-05,
      "loss": 1.0769,
      "step": 8120
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.2036825141861836e-05,
      "loss": 1.1122,
      "step": 8130
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.2010472473296895e-05,
      "loss": 1.1512,
      "step": 8140
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.198408456214783e-05,
      "loss": 1.1313,
      "step": 8150
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.195766146308555e-05,
      "loss": 1.1584,
      "step": 8160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.1931203230853874e-05,
      "loss": 1.1451,
      "step": 8170
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.190470992026944e-05,
      "loss": 1.0846,
      "step": 8180
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.18781815862215e-05,
      "loss": 1.1398,
      "step": 8190
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.1851618283671924e-05,
      "loss": 1.0838,
      "step": 8200
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.182502006765502e-05,
      "loss": 1.114,
      "step": 8210
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.179838699327739e-05,
      "loss": 1.1078,
      "step": 8220
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.1771719115717925e-05,
      "loss": 1.1787,
      "step": 8230
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.174501649022756e-05,
      "loss": 1.0983,
      "step": 8240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.171827917212926e-05,
      "loss": 1.1255,
      "step": 8250
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.1691507216817834e-05,
      "loss": 1.1437,
      "step": 8260
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.1664700679759886e-05,
      "loss": 1.1051,
      "step": 8270
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.163785961649364e-05,
      "loss": 1.1018,
      "step": 8280
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.161098408262888e-05,
      "loss": 1.129,
      "step": 8290
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.158407413384678e-05,
      "loss": 1.0712,
      "step": 8300
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.155712982589983e-05,
      "loss": 1.1564,
      "step": 8310
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.15301512146117e-05,
      "loss": 1.1427,
      "step": 8320
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.1503138355877135e-05,
      "loss": 1.0921,
      "step": 8330
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.1476091305661823e-05,
      "loss": 1.0976,
      "step": 8340
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.1449010120002305e-05,
      "loss": 1.0852,
      "step": 8350
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.142189485500585e-05,
      "loss": 1.156,
      "step": 8360
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.139474556685031e-05,
      "loss": 1.0868,
      "step": 8370
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.136756231178405e-05,
      "loss": 1.095,
      "step": 8380
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.134034514612577e-05,
      "loss": 1.1324,
      "step": 8390
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.1313094126264484e-05,
      "loss": 1.1278,
      "step": 8400
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.1285809308659296e-05,
      "loss": 1.1301,
      "step": 8410
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.125849074983936e-05,
      "loss": 1.1119,
      "step": 8420
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.1231138506403726e-05,
      "loss": 1.1334,
      "step": 8430
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.120375263502124e-05,
      "loss": 1.127,
      "step": 8440
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.117633319243039e-05,
      "loss": 1.1554,
      "step": 8450
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.114888023543927e-05,
      "loss": 1.1077,
      "step": 8460
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.112139382092537e-05,
      "loss": 1.1169,
      "step": 8470
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.10938740058355e-05,
      "loss": 1.1208,
      "step": 8480
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.106632084718568e-05,
      "loss": 1.1371,
      "step": 8490
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.103873440206101e-05,
      "loss": 1.1582,
      "step": 8500
    },
    {
      "epoch": 1.05,
      "eval_loss": 1.0412343740463257,
      "eval_runtime": 5.7639,
      "eval_samples_per_second": 89.87,
      "eval_steps_per_second": 11.277,
      "step": 8500
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.1011114727615555e-05,
      "loss": 1.1159,
      "step": 8510
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.0983461881072225e-05,
      "loss": 1.1449,
      "step": 8520
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.095577591972265e-05,
      "loss": 1.1453,
      "step": 8530
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.092805690092708e-05,
      "loss": 1.1376,
      "step": 8540
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.090030488211423e-05,
      "loss": 1.1188,
      "step": 8550
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.087251992078122e-05,
      "loss": 1.1354,
      "step": 8560
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.084470207449339e-05,
      "loss": 1.1114,
      "step": 8570
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.0816851400884226e-05,
      "loss": 1.1488,
      "step": 8580
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.078896795765522e-05,
      "loss": 1.1122,
      "step": 8590
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.076105180257577e-05,
      "loss": 1.1274,
      "step": 8600
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.073310299348304e-05,
      "loss": 1.1196,
      "step": 8610
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.070512158828184e-05,
      "loss": 1.0901,
      "step": 8620
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.06771076449445e-05,
      "loss": 1.1413,
      "step": 8630
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.064906122151081e-05,
      "loss": 1.1208,
      "step": 8640
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.06209823760878e-05,
      "loss": 1.1584,
      "step": 8650
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.059287116684972e-05,
      "loss": 1.111,
      "step": 8660
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.056472765203782e-05,
      "loss": 1.1123,
      "step": 8670
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.0536551889960326e-05,
      "loss": 1.1291,
      "step": 8680
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.050834393899227e-05,
      "loss": 1.0982,
      "step": 8690
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.048010385757534e-05,
      "loss": 1.1267,
      "step": 8700
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.0451831704217815e-05,
      "loss": 1.1009,
      "step": 8710
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.0423527537494424e-05,
      "loss": 1.112,
      "step": 8720
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.039519141604622e-05,
      "loss": 1.1231,
      "step": 8730
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.036682339858046e-05,
      "loss": 1.1602,
      "step": 8740
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.033842354387047e-05,
      "loss": 1.1013,
      "step": 8750
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.030999191075556e-05,
      "loss": 1.1679,
      "step": 8760
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.028152855814087e-05,
      "loss": 1.1438,
      "step": 8770
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.025303354499724e-05,
      "loss": 1.1588,
      "step": 8780
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.022450693036114e-05,
      "loss": 1.1448,
      "step": 8790
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.019594877333447e-05,
      "loss": 1.126,
      "step": 8800
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.016735913308452e-05,
      "loss": 1.133,
      "step": 8810
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.013873806884378e-05,
      "loss": 1.1431,
      "step": 8820
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.011008563990986e-05,
      "loss": 1.1015,
      "step": 8830
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.008140190564533e-05,
      "loss": 1.1413,
      "step": 8840
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.0052686925477656e-05,
      "loss": 1.0785,
      "step": 8850
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.0023940758899004e-05,
      "loss": 1.1091,
      "step": 8860
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.999516346546618e-05,
      "loss": 1.1731,
      "step": 8870
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.996635510480046e-05,
      "loss": 1.0999,
      "step": 8880
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.993751573658749e-05,
      "loss": 1.1424,
      "step": 8890
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.990864542057716e-05,
      "loss": 1.117,
      "step": 8900
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.987974421658348e-05,
      "loss": 1.115,
      "step": 8910
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.9850812184484454e-05,
      "loss": 1.1463,
      "step": 8920
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.982184938422196e-05,
      "loss": 1.0894,
      "step": 8930
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.97928558758016e-05,
      "loss": 1.1138,
      "step": 8940
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.976383171929263e-05,
      "loss": 1.0911,
      "step": 8950
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.973477697482778e-05,
      "loss": 1.1567,
      "step": 8960
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.970569170260317e-05,
      "loss": 1.1079,
      "step": 8970
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.967657596287814e-05,
      "loss": 1.098,
      "step": 8980
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.964742981597518e-05,
      "loss": 1.1392,
      "step": 8990
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.961825332227978e-05,
      "loss": 1.1326,
      "step": 9000
    },
    {
      "epoch": 1.11,
      "eval_loss": 1.0326589345932007,
      "eval_runtime": 5.7632,
      "eval_samples_per_second": 89.881,
      "eval_steps_per_second": 11.279,
      "step": 9000
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.958904654224028e-05,
      "loss": 1.1302,
      "step": 9010
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.955980953636779e-05,
      "loss": 1.1232,
      "step": 9020
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.953054236523601e-05,
      "loss": 1.1298,
      "step": 9030
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.9501245089481194e-05,
      "loss": 1.1508,
      "step": 9040
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.94719177698019e-05,
      "loss": 1.1424,
      "step": 9050
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.944256046695899e-05,
      "loss": 1.0716,
      "step": 9060
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.94131732417754e-05,
      "loss": 1.1394,
      "step": 9070
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.93837561551361e-05,
      "loss": 1.1057,
      "step": 9080
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.935430926798788e-05,
      "loss": 1.1101,
      "step": 9090
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.932483264133932e-05,
      "loss": 1.1229,
      "step": 9100
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.929532633626058e-05,
      "loss": 1.1618,
      "step": 9110
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.926579041388333e-05,
      "loss": 1.1355,
      "step": 9120
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.923622493540059e-05,
      "loss": 1.0898,
      "step": 9130
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.920662996206661e-05,
      "loss": 1.1455,
      "step": 9140
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.917700555519677e-05,
      "loss": 1.156,
      "step": 9150
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.914735177616741e-05,
      "loss": 1.119,
      "step": 9160
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.911766868641573e-05,
      "loss": 1.1109,
      "step": 9170
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.908795634743965e-05,
      "loss": 1.1524,
      "step": 9180
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.90582148207977e-05,
      "loss": 1.1376,
      "step": 9190
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.902844416810888e-05,
      "loss": 1.1177,
      "step": 9200
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.8998644451052536e-05,
      "loss": 1.1328,
      "step": 9210
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.896881573136821e-05,
      "loss": 1.1206,
      "step": 9220
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.8938958070855566e-05,
      "loss": 1.1143,
      "step": 9230
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.8909071531374184e-05,
      "loss": 1.0996,
      "step": 9240
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.8879156174843536e-05,
      "loss": 1.1494,
      "step": 9250
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.8849212063242726e-05,
      "loss": 1.1447,
      "step": 9260
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.88192392586105e-05,
      "loss": 1.1533,
      "step": 9270
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.878923782304499e-05,
      "loss": 1.0953,
      "step": 9280
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.8759207818703706e-05,
      "loss": 1.1402,
      "step": 9290
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.872914930780331e-05,
      "loss": 1.1039,
      "step": 9300
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.8699062352619524e-05,
      "loss": 1.1293,
      "step": 9310
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.8668947015487015e-05,
      "loss": 1.1523,
      "step": 9320
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.863880335879925e-05,
      "loss": 1.1277,
      "step": 9330
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.860863144500836e-05,
      "loss": 1.1138,
      "step": 9340
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.857843133662503e-05,
      "loss": 1.1411,
      "step": 9350
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.854820309621836e-05,
      "loss": 1.0831,
      "step": 9360
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.851794678641572e-05,
      "loss": 1.0936,
      "step": 9370
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.848766246990264e-05,
      "loss": 1.1398,
      "step": 9380
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.845735020942268e-05,
      "loss": 1.0537,
      "step": 9390
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.842701006777731e-05,
      "loss": 1.1364,
      "step": 9400
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.839664210782573e-05,
      "loss": 1.1252,
      "step": 9410
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.8366246392484796e-05,
      "loss": 1.0967,
      "step": 9420
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.8335822984728866e-05,
      "loss": 1.077,
      "step": 9430
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.830537194758967e-05,
      "loss": 1.1082,
      "step": 9440
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.827489334415618e-05,
      "loss": 1.1591,
      "step": 9450
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.824438723757447e-05,
      "loss": 1.1019,
      "step": 9460
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.8213853691047644e-05,
      "loss": 1.1382,
      "step": 9470
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.818329276783558e-05,
      "loss": 1.1209,
      "step": 9480
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.815270453125493e-05,
      "loss": 1.1131,
      "step": 9490
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.8122089044678916e-05,
      "loss": 1.1313,
      "step": 9500
    },
    {
      "epoch": 1.18,
      "eval_loss": 1.0298551321029663,
      "eval_runtime": 5.7612,
      "eval_samples_per_second": 89.911,
      "eval_steps_per_second": 11.282,
      "step": 9500
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.809144637153722e-05,
      "loss": 1.1617,
      "step": 9510
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.8060776575315865e-05,
      "loss": 1.101,
      "step": 9520
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.8030079719557026e-05,
      "loss": 1.1377,
      "step": 9530
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.7999355867858985e-05,
      "loss": 1.1324,
      "step": 9540
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.796860508387593e-05,
      "loss": 1.1266,
      "step": 9550
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.7937827431317866e-05,
      "loss": 1.0918,
      "step": 9560
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.790702297395044e-05,
      "loss": 1.0917,
      "step": 9570
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.787619177559485e-05,
      "loss": 1.1262,
      "step": 9580
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.784533390012769e-05,
      "loss": 1.0977,
      "step": 9590
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.7814449411480836e-05,
      "loss": 1.1104,
      "step": 9600
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.7783538373641296e-05,
      "loss": 1.0976,
      "step": 9610
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.775260085065107e-05,
      "loss": 1.1409,
      "step": 9620
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.772163690660704e-05,
      "loss": 1.1272,
      "step": 9630
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.7690646605660826e-05,
      "loss": 1.1082,
      "step": 9640
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.7659630012018676e-05,
      "loss": 1.0926,
      "step": 9650
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.7628587189941276e-05,
      "loss": 1.1474,
      "step": 9660
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.759751820374367e-05,
      "loss": 1.0877,
      "step": 9670
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.756642311779511e-05,
      "loss": 1.1289,
      "step": 9680
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.753530199651892e-05,
      "loss": 1.101,
      "step": 9690
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.750415490439237e-05,
      "loss": 1.0849,
      "step": 9700
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.7472981905946535e-05,
      "loss": 1.087,
      "step": 9710
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.744178306576614e-05,
      "loss": 1.1201,
      "step": 9720
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.741055844848949e-05,
      "loss": 1.1066,
      "step": 9730
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.737930811880827e-05,
      "loss": 1.0818,
      "step": 9740
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.7348032141467414e-05,
      "loss": 1.1116,
      "step": 9750
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.7316730581265054e-05,
      "loss": 1.1193,
      "step": 9760
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.728540350305227e-05,
      "loss": 1.0784,
      "step": 9770
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.725405097173306e-05,
      "loss": 1.1277,
      "step": 9780
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.722267305226409e-05,
      "loss": 1.1043,
      "step": 9790
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.71912698096547e-05,
      "loss": 1.1234,
      "step": 9800
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.7159841308966626e-05,
      "loss": 1.1158,
      "step": 9810
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.7128387615313986e-05,
      "loss": 1.1216,
      "step": 9820
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.709690879386306e-05,
      "loss": 1.0826,
      "step": 9830
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.7065404909832224e-05,
      "loss": 1.1404,
      "step": 9840
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.7033876028491734e-05,
      "loss": 1.1411,
      "step": 9850
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.700232221516367e-05,
      "loss": 1.0914,
      "step": 9860
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.6970743535221755e-05,
      "loss": 1.0979,
      "step": 9870
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.693914005409123e-05,
      "loss": 1.1405,
      "step": 9880
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.690751183724872e-05,
      "loss": 1.0844,
      "step": 9890
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.6875858950222095e-05,
      "loss": 1.0991,
      "step": 9900
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.684418145859036e-05,
      "loss": 1.0847,
      "step": 9910
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.681247942798345e-05,
      "loss": 1.1008,
      "step": 9920
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.6780752924082175e-05,
      "loss": 1.1674,
      "step": 9930
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.6749002012618055e-05,
      "loss": 1.0449,
      "step": 9940
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.671722675937316e-05,
      "loss": 1.1063,
      "step": 9950
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.6685427230179984e-05,
      "loss": 1.1387,
      "step": 9960
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.6653603490921335e-05,
      "loss": 1.1268,
      "step": 9970
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.662175560753018e-05,
      "loss": 1.1255,
      "step": 9980
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.6589883645989495e-05,
      "loss": 1.1349,
      "step": 9990
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.655798767233215e-05,
      "loss": 1.1007,
      "step": 10000
    },
    {
      "epoch": 1.24,
      "eval_loss": 1.0268759727478027,
      "eval_runtime": 5.7616,
      "eval_samples_per_second": 89.906,
      "eval_steps_per_second": 11.282,
      "step": 10000
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.652606775264077e-05,
      "loss": 1.1228,
      "step": 10010
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.649412395304757e-05,
      "loss": 1.1163,
      "step": 10020
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.646215633973425e-05,
      "loss": 1.1201,
      "step": 10030
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.643016497893187e-05,
      "loss": 1.1287,
      "step": 10040
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.639814993692065e-05,
      "loss": 1.115,
      "step": 10050
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.6366111280029927e-05,
      "loss": 1.1101,
      "step": 10060
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.6334049074637896e-05,
      "loss": 1.1096,
      "step": 10070
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.63019633871716e-05,
      "loss": 1.0797,
      "step": 10080
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.6269854284106686e-05,
      "loss": 1.08,
      "step": 10090
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.623772183196735e-05,
      "loss": 1.1361,
      "step": 10100
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.620556609732616e-05,
      "loss": 1.1033,
      "step": 10110
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.617338714680389e-05,
      "loss": 1.1069,
      "step": 10120
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.614118504706943e-05,
      "loss": 1.1182,
      "step": 10130
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.610895986483965e-05,
      "loss": 1.1513,
      "step": 10140
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.6076711666879234e-05,
      "loss": 1.1092,
      "step": 10150
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.604444052000052e-05,
      "loss": 1.1428,
      "step": 10160
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.601214649106343e-05,
      "loss": 1.1113,
      "step": 10170
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.597982964697527e-05,
      "loss": 1.1404,
      "step": 10180
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.594749005469063e-05,
      "loss": 1.1082,
      "step": 10190
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.5915127781211215e-05,
      "loss": 1.1014,
      "step": 10200
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.5882742893585745e-05,
      "loss": 1.1115,
      "step": 10210
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.585033545890976e-05,
      "loss": 1.1311,
      "step": 10220
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.581790554432554e-05,
      "loss": 1.0668,
      "step": 10230
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.5785453217021926e-05,
      "loss": 1.1143,
      "step": 10240
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.575297854423421e-05,
      "loss": 1.1524,
      "step": 10250
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.572048159324394e-05,
      "loss": 1.1086,
      "step": 10260
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.5687962431378875e-05,
      "loss": 1.0636,
      "step": 10270
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.565542112601275e-05,
      "loss": 1.1111,
      "step": 10280
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.56228577445652e-05,
      "loss": 1.1441,
      "step": 10290
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.559027235450158e-05,
      "loss": 1.1377,
      "step": 10300
    },
    {
      "epoch": 1.28,
      "learning_rate": 3.555766502333285e-05,
      "loss": 1.1123,
      "step": 10310
    },
    {
      "epoch": 1.28,
      "learning_rate": 3.552503581861544e-05,
      "loss": 1.1103,
      "step": 10320
    },
    {
      "epoch": 1.28,
      "learning_rate": 3.549238480795107e-05,
      "loss": 1.1019,
      "step": 10330
    },
    {
      "epoch": 1.28,
      "learning_rate": 3.545971205898668e-05,
      "loss": 1.153,
      "step": 10340
    },
    {
      "epoch": 1.28,
      "learning_rate": 3.5427017639414206e-05,
      "loss": 1.1124,
      "step": 10350
    },
    {
      "epoch": 1.28,
      "learning_rate": 3.53943016169705e-05,
      "loss": 1.0971,
      "step": 10360
    },
    {
      "epoch": 1.28,
      "learning_rate": 3.536156405943717e-05,
      "loss": 1.0987,
      "step": 10370
    },
    {
      "epoch": 1.28,
      "learning_rate": 3.532880503464045e-05,
      "loss": 1.104,
      "step": 10380
    },
    {
      "epoch": 1.29,
      "learning_rate": 3.529602461045102e-05,
      "loss": 1.1113,
      "step": 10390
    },
    {
      "epoch": 1.29,
      "learning_rate": 3.526322285478394e-05,
      "loss": 1.0833,
      "step": 10400
    },
    {
      "epoch": 1.29,
      "learning_rate": 3.523039983559843e-05,
      "loss": 1.0938,
      "step": 10410
    },
    {
      "epoch": 1.29,
      "learning_rate": 3.519755562089778e-05,
      "loss": 1.0916,
      "step": 10420
    },
    {
      "epoch": 1.29,
      "learning_rate": 3.516469027872917e-05,
      "loss": 1.1295,
      "step": 10430
    },
    {
      "epoch": 1.29,
      "learning_rate": 3.51318038771836e-05,
      "loss": 1.1001,
      "step": 10440
    },
    {
      "epoch": 1.29,
      "learning_rate": 3.509889648439565e-05,
      "loss": 1.1012,
      "step": 10450
    },
    {
      "epoch": 1.29,
      "learning_rate": 3.506596816854343e-05,
      "loss": 1.0968,
      "step": 10460
    },
    {
      "epoch": 1.3,
      "learning_rate": 3.503301899784837e-05,
      "loss": 1.147,
      "step": 10470
    },
    {
      "epoch": 1.3,
      "learning_rate": 3.500004904057511e-05,
      "loss": 1.1021,
      "step": 10480
    },
    {
      "epoch": 1.3,
      "learning_rate": 3.496705836503138e-05,
      "loss": 1.1351,
      "step": 10490
    },
    {
      "epoch": 1.3,
      "learning_rate": 3.4934047039567806e-05,
      "loss": 1.0983,
      "step": 10500
    },
    {
      "epoch": 1.3,
      "eval_loss": 1.0228146314620972,
      "eval_runtime": 5.7633,
      "eval_samples_per_second": 89.88,
      "eval_steps_per_second": 11.278,
      "step": 10500
    },
    {
      "epoch": 1.3,
      "learning_rate": 3.4901015132577825e-05,
      "loss": 1.0657,
      "step": 10510
    },
    {
      "epoch": 1.3,
      "learning_rate": 3.4867962712497486e-05,
      "loss": 1.111,
      "step": 10520
    },
    {
      "epoch": 1.3,
      "learning_rate": 3.483488984780536e-05,
      "loss": 1.0828,
      "step": 10530
    },
    {
      "epoch": 1.3,
      "learning_rate": 3.480179660702235e-05,
      "loss": 1.1044,
      "step": 10540
    },
    {
      "epoch": 1.31,
      "learning_rate": 3.476868305871162e-05,
      "loss": 1.0812,
      "step": 10550
    },
    {
      "epoch": 1.31,
      "learning_rate": 3.473554927147835e-05,
      "loss": 1.1466,
      "step": 10560
    },
    {
      "epoch": 1.31,
      "learning_rate": 3.470239531396969e-05,
      "loss": 1.0785,
      "step": 10570
    },
    {
      "epoch": 1.31,
      "learning_rate": 3.4669221254874586e-05,
      "loss": 1.1019,
      "step": 10580
    },
    {
      "epoch": 1.31,
      "learning_rate": 3.463602716292358e-05,
      "loss": 1.1121,
      "step": 10590
    },
    {
      "epoch": 1.31,
      "learning_rate": 3.4602813106888784e-05,
      "loss": 1.1051,
      "step": 10600
    },
    {
      "epoch": 1.31,
      "learning_rate": 3.456957915558362e-05,
      "loss": 1.1127,
      "step": 10610
    },
    {
      "epoch": 1.31,
      "learning_rate": 3.4536325377862765e-05,
      "loss": 1.151,
      "step": 10620
    },
    {
      "epoch": 1.31,
      "learning_rate": 3.450305184262194e-05,
      "loss": 1.1116,
      "step": 10630
    },
    {
      "epoch": 1.32,
      "learning_rate": 3.446975861879783e-05,
      "loss": 1.1014,
      "step": 10640
    },
    {
      "epoch": 1.32,
      "learning_rate": 3.44364457753679e-05,
      "loss": 1.1474,
      "step": 10650
    },
    {
      "epoch": 1.32,
      "learning_rate": 3.440311338135025e-05,
      "loss": 1.1454,
      "step": 10660
    },
    {
      "epoch": 1.32,
      "learning_rate": 3.43697615058035e-05,
      "loss": 1.1614,
      "step": 10670
    },
    {
      "epoch": 1.32,
      "learning_rate": 3.433639021782664e-05,
      "loss": 1.1296,
      "step": 10680
    },
    {
      "epoch": 1.32,
      "learning_rate": 3.4302999586558855e-05,
      "loss": 1.132,
      "step": 10690
    },
    {
      "epoch": 1.32,
      "learning_rate": 3.426958968117943e-05,
      "loss": 1.1189,
      "step": 10700
    },
    {
      "epoch": 1.32,
      "learning_rate": 3.423616057090755e-05,
      "loss": 1.0983,
      "step": 10710
    },
    {
      "epoch": 1.33,
      "learning_rate": 3.420271232500224e-05,
      "loss": 1.1026,
      "step": 10720
    },
    {
      "epoch": 1.33,
      "learning_rate": 3.416924501276212e-05,
      "loss": 1.1262,
      "step": 10730
    },
    {
      "epoch": 1.33,
      "learning_rate": 3.4135758703525337e-05,
      "loss": 1.1233,
      "step": 10740
    },
    {
      "epoch": 1.33,
      "learning_rate": 3.41022534666694e-05,
      "loss": 1.0908,
      "step": 10750
    },
    {
      "epoch": 1.33,
      "learning_rate": 3.406872937161101e-05,
      "loss": 1.0839,
      "step": 10760
    },
    {
      "epoch": 1.33,
      "learning_rate": 3.403518648780597e-05,
      "loss": 1.1074,
      "step": 10770
    },
    {
      "epoch": 1.33,
      "learning_rate": 3.4001624884748975e-05,
      "loss": 1.0915,
      "step": 10780
    },
    {
      "epoch": 1.33,
      "learning_rate": 3.396804463197353e-05,
      "loss": 1.1377,
      "step": 10790
    },
    {
      "epoch": 1.34,
      "learning_rate": 3.393444579905177e-05,
      "loss": 1.0913,
      "step": 10800
    },
    {
      "epoch": 1.34,
      "learning_rate": 3.390082845559434e-05,
      "loss": 1.1673,
      "step": 10810
    },
    {
      "epoch": 1.34,
      "learning_rate": 3.386719267125018e-05,
      "loss": 1.12,
      "step": 10820
    },
    {
      "epoch": 1.34,
      "learning_rate": 3.383353851570652e-05,
      "loss": 1.0588,
      "step": 10830
    },
    {
      "epoch": 1.34,
      "learning_rate": 3.379986605868858e-05,
      "loss": 1.0715,
      "step": 10840
    },
    {
      "epoch": 1.34,
      "learning_rate": 3.3766175369959536e-05,
      "loss": 1.0724,
      "step": 10850
    },
    {
      "epoch": 1.34,
      "learning_rate": 3.373246651932032e-05,
      "loss": 1.0968,
      "step": 10860
    },
    {
      "epoch": 1.34,
      "learning_rate": 3.36987395766095e-05,
      "loss": 1.092,
      "step": 10870
    },
    {
      "epoch": 1.35,
      "learning_rate": 3.366499461170314e-05,
      "loss": 1.1112,
      "step": 10880
    },
    {
      "epoch": 1.35,
      "learning_rate": 3.363123169451461e-05,
      "loss": 1.0916,
      "step": 10890
    },
    {
      "epoch": 1.35,
      "learning_rate": 3.359745089499448e-05,
      "loss": 1.1086,
      "step": 10900
    },
    {
      "epoch": 1.35,
      "learning_rate": 3.356365228313041e-05,
      "loss": 1.0841,
      "step": 10910
    },
    {
      "epoch": 1.35,
      "learning_rate": 3.352983592894692e-05,
      "loss": 1.0953,
      "step": 10920
    },
    {
      "epoch": 1.35,
      "learning_rate": 3.349600190250531e-05,
      "loss": 1.0618,
      "step": 10930
    },
    {
      "epoch": 1.35,
      "learning_rate": 3.346215027390349e-05,
      "loss": 1.065,
      "step": 10940
    },
    {
      "epoch": 1.35,
      "learning_rate": 3.342828111327582e-05,
      "loss": 1.1423,
      "step": 10950
    },
    {
      "epoch": 1.36,
      "learning_rate": 3.3394394490793004e-05,
      "loss": 1.0813,
      "step": 10960
    },
    {
      "epoch": 1.36,
      "learning_rate": 3.336049047666193e-05,
      "loss": 1.1,
      "step": 10970
    },
    {
      "epoch": 1.36,
      "learning_rate": 3.3326569141125497e-05,
      "loss": 1.1031,
      "step": 10980
    },
    {
      "epoch": 1.36,
      "learning_rate": 3.3292630554462505e-05,
      "loss": 1.0862,
      "step": 10990
    },
    {
      "epoch": 1.36,
      "learning_rate": 3.3258674786987496e-05,
      "loss": 1.0946,
      "step": 11000
    },
    {
      "epoch": 1.36,
      "eval_loss": 1.0201221704483032,
      "eval_runtime": 5.7649,
      "eval_samples_per_second": 89.855,
      "eval_steps_per_second": 11.275,
      "step": 11000
    },
    {
      "epoch": 1.36,
      "learning_rate": 3.322470190905059e-05,
      "loss": 1.0983,
      "step": 11010
    },
    {
      "epoch": 1.36,
      "learning_rate": 3.319071199103737e-05,
      "loss": 1.1614,
      "step": 11020
    },
    {
      "epoch": 1.36,
      "learning_rate": 3.315670510336874e-05,
      "loss": 1.1385,
      "step": 11030
    },
    {
      "epoch": 1.37,
      "learning_rate": 3.312268131650072e-05,
      "loss": 1.0788,
      "step": 11040
    },
    {
      "epoch": 1.37,
      "learning_rate": 3.308864070092438e-05,
      "loss": 1.1204,
      "step": 11050
    },
    {
      "epoch": 1.37,
      "learning_rate": 3.305458332716565e-05,
      "loss": 1.1076,
      "step": 11060
    },
    {
      "epoch": 1.37,
      "learning_rate": 3.3020509265785154e-05,
      "loss": 1.08,
      "step": 11070
    },
    {
      "epoch": 1.37,
      "learning_rate": 3.2986418587378124e-05,
      "loss": 1.118,
      "step": 11080
    },
    {
      "epoch": 1.37,
      "learning_rate": 3.2952311362574184e-05,
      "loss": 1.1275,
      "step": 11090
    },
    {
      "epoch": 1.37,
      "learning_rate": 3.2918187662037276e-05,
      "loss": 1.1198,
      "step": 11100
    },
    {
      "epoch": 1.37,
      "learning_rate": 3.288404755646545e-05,
      "loss": 1.1093,
      "step": 11110
    },
    {
      "epoch": 1.38,
      "learning_rate": 3.284989111659076e-05,
      "loss": 1.116,
      "step": 11120
    },
    {
      "epoch": 1.38,
      "learning_rate": 3.281571841317909e-05,
      "loss": 1.1058,
      "step": 11130
    },
    {
      "epoch": 1.38,
      "learning_rate": 3.2781529517030026e-05,
      "loss": 1.1132,
      "step": 11140
    },
    {
      "epoch": 1.38,
      "learning_rate": 3.27473244989767e-05,
      "loss": 1.1322,
      "step": 11150
    },
    {
      "epoch": 1.38,
      "learning_rate": 3.271310342988564e-05,
      "loss": 1.0679,
      "step": 11160
    },
    {
      "epoch": 1.38,
      "learning_rate": 3.267886638065664e-05,
      "loss": 1.1059,
      "step": 11170
    },
    {
      "epoch": 1.38,
      "learning_rate": 3.26446134222226e-05,
      "loss": 1.1099,
      "step": 11180
    },
    {
      "epoch": 1.38,
      "learning_rate": 3.261034462554937e-05,
      "loss": 1.1102,
      "step": 11190
    },
    {
      "epoch": 1.39,
      "learning_rate": 3.257606006163563e-05,
      "loss": 1.1212,
      "step": 11200
    },
    {
      "epoch": 1.39,
      "learning_rate": 3.2541759801512714e-05,
      "loss": 1.0798,
      "step": 11210
    },
    {
      "epoch": 1.39,
      "learning_rate": 3.250744391624448e-05,
      "loss": 1.0891,
      "step": 11220
    },
    {
      "epoch": 1.39,
      "learning_rate": 3.247311247692717e-05,
      "loss": 1.0806,
      "step": 11230
    },
    {
      "epoch": 1.39,
      "learning_rate": 3.243876555468922e-05,
      "loss": 1.0772,
      "step": 11240
    },
    {
      "epoch": 1.39,
      "learning_rate": 3.2404403220691174e-05,
      "loss": 1.1413,
      "step": 11250
    },
    {
      "epoch": 1.39,
      "learning_rate": 3.2370025546125494e-05,
      "loss": 1.1231,
      "step": 11260
    },
    {
      "epoch": 1.39,
      "learning_rate": 3.2335632602216434e-05,
      "loss": 1.0717,
      "step": 11270
    },
    {
      "epoch": 1.4,
      "learning_rate": 3.230122446021986e-05,
      "loss": 1.0537,
      "step": 11280
    },
    {
      "epoch": 1.4,
      "learning_rate": 3.2266801191423155e-05,
      "loss": 1.1031,
      "step": 11290
    },
    {
      "epoch": 1.4,
      "learning_rate": 3.223236286714503e-05,
      "loss": 1.0522,
      "step": 11300
    },
    {
      "epoch": 1.4,
      "learning_rate": 3.2197909558735373e-05,
      "loss": 1.1081,
      "step": 11310
    },
    {
      "epoch": 1.4,
      "learning_rate": 3.216344133757514e-05,
      "loss": 1.129,
      "step": 11320
    },
    {
      "epoch": 1.4,
      "learning_rate": 3.212895827507617e-05,
      "loss": 1.0907,
      "step": 11330
    },
    {
      "epoch": 1.4,
      "learning_rate": 3.209446044268106e-05,
      "loss": 1.0669,
      "step": 11340
    },
    {
      "epoch": 1.4,
      "learning_rate": 3.2059947911863e-05,
      "loss": 1.1227,
      "step": 11350
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.202542075412563e-05,
      "loss": 1.1408,
      "step": 11360
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.199087904100291e-05,
      "loss": 1.0806,
      "step": 11370
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.195632284405893e-05,
      "loss": 1.101,
      "step": 11380
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.1921752234887805e-05,
      "loss": 1.1398,
      "step": 11390
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.1887167285113504e-05,
      "loss": 1.1293,
      "step": 11400
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.1852568066389724e-05,
      "loss": 1.0837,
      "step": 11410
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.1817954650399704e-05,
      "loss": 1.0671,
      "step": 11420
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.17833271088561e-05,
      "loss": 1.1242,
      "step": 11430
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.1748685513500845e-05,
      "loss": 1.0606,
      "step": 11440
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.171402993610496e-05,
      "loss": 1.0741,
      "step": 11450
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.167936044846849e-05,
      "loss": 1.0893,
      "step": 11460
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.164467712242023e-05,
      "loss": 1.1121,
      "step": 11470
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.160998002981771e-05,
      "loss": 1.0779,
      "step": 11480
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.1575269242546927e-05,
      "loss": 1.0568,
      "step": 11490
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.15405448325223e-05,
      "loss": 1.127,
      "step": 11500
    },
    {
      "epoch": 1.42,
      "eval_loss": 1.0143646001815796,
      "eval_runtime": 5.7627,
      "eval_samples_per_second": 89.888,
      "eval_steps_per_second": 11.279,
      "step": 11500
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.1505806871686426e-05,
      "loss": 1.1001,
      "step": 11510
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.147105543201002e-05,
      "loss": 1.1072,
      "step": 11520
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.1436290585491686e-05,
      "loss": 1.1027,
      "step": 11530
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.140151240415782e-05,
      "loss": 1.0892,
      "step": 11540
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.136672096006246e-05,
      "loss": 1.1202,
      "step": 11550
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.1331916325287094e-05,
      "loss": 1.0918,
      "step": 11560
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.1297098571940556e-05,
      "loss": 1.0614,
      "step": 11570
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.126226777215886e-05,
      "loss": 1.1036,
      "step": 11580
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.122742399810503e-05,
      "loss": 1.0841,
      "step": 11590
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.1192567321969e-05,
      "loss": 1.0612,
      "step": 11600
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.1157697815967424e-05,
      "loss": 1.1428,
      "step": 11610
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.112281555234352e-05,
      "loss": 1.0921,
      "step": 11620
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.108792060336695e-05,
      "loss": 1.0796,
      "step": 11630
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.105301304133368e-05,
      "loss": 1.0947,
      "step": 11640
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.1018092938565765e-05,
      "loss": 1.1185,
      "step": 11650
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.098316036741127e-05,
      "loss": 1.0856,
      "step": 11660
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.0948215400244085e-05,
      "loss": 1.1219,
      "step": 11670
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.091325810946378e-05,
      "loss": 1.0856,
      "step": 11680
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.087828856749547e-05,
      "loss": 1.0639,
      "step": 11690
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.084330684678964e-05,
      "loss": 1.0643,
      "step": 11700
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.0808313019822e-05,
      "loss": 1.1239,
      "step": 11710
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.0773307159093365e-05,
      "loss": 1.0759,
      "step": 11720
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.0738289337129454e-05,
      "loss": 1.0969,
      "step": 11730
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.0703259626480797e-05,
      "loss": 1.1094,
      "step": 11740
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.066821809972253e-05,
      "loss": 1.0912,
      "step": 11750
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.063316482945429e-05,
      "loss": 1.1278,
      "step": 11760
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.059809988830002e-05,
      "loss": 1.1022,
      "step": 11770
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.056302334890786e-05,
      "loss": 1.1197,
      "step": 11780
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.0527935283949986e-05,
      "loss": 1.1254,
      "step": 11790
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.0492835766122442e-05,
      "loss": 1.1088,
      "step": 11800
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.0457724868145e-05,
      "loss": 1.0983,
      "step": 11810
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.0422602662761002e-05,
      "loss": 1.0614,
      "step": 11820
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.0387469222737236e-05,
      "loss": 1.116,
      "step": 11830
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.035232462086376e-05,
      "loss": 1.0658,
      "step": 11840
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.0317168929953743e-05,
      "loss": 1.0905,
      "step": 11850
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.028200222284334e-05,
      "loss": 1.0912,
      "step": 11860
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.0246824572391542e-05,
      "loss": 1.0641,
      "step": 11870
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.0211636051479985e-05,
      "loss": 1.1061,
      "step": 11880
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.0176436733012846e-05,
      "loss": 1.1396,
      "step": 11890
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.0141226689916667e-05,
      "loss": 1.0759,
      "step": 11900
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.0106005995140212e-05,
      "loss": 1.1406,
      "step": 11910
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.0070774721654306e-05,
      "loss": 1.1354,
      "step": 11920
    },
    {
      "epoch": 1.48,
      "learning_rate": 3.0035532942451705e-05,
      "loss": 1.1265,
      "step": 11930
    },
    {
      "epoch": 1.48,
      "learning_rate": 3.0000280730546913e-05,
      "loss": 1.0837,
      "step": 11940
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.9965018158976064e-05,
      "loss": 1.1294,
      "step": 11950
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.9929745300796747e-05,
      "loss": 1.0733,
      "step": 11960
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.9894462229087866e-05,
      "loss": 1.1176,
      "step": 11970
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.9859169016949484e-05,
      "loss": 1.1409,
      "step": 11980
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.982386573750267e-05,
      "loss": 1.1176,
      "step": 11990
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.9788552463889364e-05,
      "loss": 1.1087,
      "step": 12000
    },
    {
      "epoch": 1.48,
      "eval_loss": 1.014037847518921,
      "eval_runtime": 5.7696,
      "eval_samples_per_second": 89.781,
      "eval_steps_per_second": 11.266,
      "step": 12000
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.97532292692722e-05,
      "loss": 1.1168,
      "step": 12010
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.9717896226834364e-05,
      "loss": 1.0966,
      "step": 12020
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.9682553409779447e-05,
      "loss": 1.1,
      "step": 12030
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.9647200891331306e-05,
      "loss": 1.0777,
      "step": 12040
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.961183874473388e-05,
      "loss": 1.0779,
      "step": 12050
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.9576467043251054e-05,
      "loss": 1.1108,
      "step": 12060
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.9541085860166528e-05,
      "loss": 1.1019,
      "step": 12070
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.950569526878363e-05,
      "loss": 1.1045,
      "step": 12080
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.9470295342425186e-05,
      "loss": 1.1158,
      "step": 12090
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.943488615443336e-05,
      "loss": 1.1094,
      "step": 12100
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.9399467778169502e-05,
      "loss": 1.1158,
      "step": 12110
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.9364040287014012e-05,
      "loss": 1.0872,
      "step": 12120
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.932860375436615e-05,
      "loss": 1.0977,
      "step": 12130
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.9293158253643936e-05,
      "loss": 1.0815,
      "step": 12140
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.9257703858283958e-05,
      "loss": 1.087,
      "step": 12150
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.9222240641741216e-05,
      "loss": 1.1078,
      "step": 12160
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.918676867748902e-05,
      "loss": 1.0413,
      "step": 12170
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.9151288039018766e-05,
      "loss": 1.1274,
      "step": 12180
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.911579879983985e-05,
      "loss": 1.0624,
      "step": 12190
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.9080301033479475e-05,
      "loss": 1.0967,
      "step": 12200
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.9044794813482518e-05,
      "loss": 1.0794,
      "step": 12210
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.9009280213411356e-05,
      "loss": 1.0939,
      "step": 12220
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.8973757306845738e-05,
      "loss": 1.1248,
      "step": 12230
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.8938226167382636e-05,
      "loss": 1.0969,
      "step": 12240
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.8902686868636047e-05,
      "loss": 1.1171,
      "step": 12250
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.8867139484236898e-05,
      "loss": 1.0909,
      "step": 12260
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.8831584087832864e-05,
      "loss": 1.0625,
      "step": 12270
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.879602075308821e-05,
      "loss": 1.0956,
      "step": 12280
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.8760449553683656e-05,
      "loss": 1.093,
      "step": 12290
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.872487056331621e-05,
      "loss": 1.0674,
      "step": 12300
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.868928385569903e-05,
      "loss": 1.0648,
      "step": 12310
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.8653689504561253e-05,
      "loss": 1.1238,
      "step": 12320
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.8618087583647857e-05,
      "loss": 1.12,
      "step": 12330
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.8582478166719512e-05,
      "loss": 1.115,
      "step": 12340
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.85468613275524e-05,
      "loss": 1.1158,
      "step": 12350
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.851123713993809e-05,
      "loss": 1.1165,
      "step": 12360
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.8475605677683383e-05,
      "loss": 1.0985,
      "step": 12370
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.8439967014610134e-05,
      "loss": 1.0928,
      "step": 12380
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.8404321224555135e-05,
      "loss": 1.0878,
      "step": 12390
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.8368668381369934e-05,
      "loss": 1.0734,
      "step": 12400
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.8333008558920694e-05,
      "loss": 1.1231,
      "step": 12410
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.829734183108803e-05,
      "loss": 1.0822,
      "step": 12420
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.8261668271766882e-05,
      "loss": 1.0886,
      "step": 12430
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.8225987954866325e-05,
      "loss": 1.0877,
      "step": 12440
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.8190300954309447e-05,
      "loss": 1.0806,
      "step": 12450
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.8154607344033174e-05,
      "loss": 1.1316,
      "step": 12460
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.8118907197988133e-05,
      "loss": 1.1096,
      "step": 12470
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.8083200590138492e-05,
      "loss": 1.0957,
      "step": 12480
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.8047487594461797e-05,
      "loss": 1.0703,
      "step": 12490
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.801176828494884e-05,
      "loss": 1.0958,
      "step": 12500
    },
    {
      "epoch": 1.55,
      "eval_loss": 1.0098438262939453,
      "eval_runtime": 5.7652,
      "eval_samples_per_second": 89.849,
      "eval_steps_per_second": 11.275,
      "step": 12500
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.7976042735603496e-05,
      "loss": 1.1187,
      "step": 12510
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.794031102044255e-05,
      "loss": 1.0942,
      "step": 12520
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.7904573213495576e-05,
      "loss": 1.1351,
      "step": 12530
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.7868829388804775e-05,
      "loss": 1.0822,
      "step": 12540
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.783307962042479e-05,
      "loss": 1.1083,
      "step": 12550
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.7797323982422614e-05,
      "loss": 1.0942,
      "step": 12560
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.776156254887736e-05,
      "loss": 1.0598,
      "step": 12570
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.7725795393880195e-05,
      "loss": 1.1061,
      "step": 12580
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.7690022591534093e-05,
      "loss": 1.0706,
      "step": 12590
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.765424421595376e-05,
      "loss": 1.0994,
      "step": 12600
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.7618460341265438e-05,
      "loss": 1.1187,
      "step": 12610
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.758267104160676e-05,
      "loss": 1.0757,
      "step": 12620
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.7546876391126608e-05,
      "loss": 1.1194,
      "step": 12630
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.7511076463984936e-05,
      "loss": 1.1067,
      "step": 12640
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.7475271334352644e-05,
      "loss": 1.0822,
      "step": 12650
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.7439461076411395e-05,
      "loss": 1.0869,
      "step": 12660
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.7403645764353498e-05,
      "loss": 1.111,
      "step": 12670
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.7367825472381707e-05,
      "loss": 1.0966,
      "step": 12680
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.7332000274709124e-05,
      "loss": 1.1024,
      "step": 12690
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.7296170245558988e-05,
      "loss": 1.1165,
      "step": 12700
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.726033545916456e-05,
      "loss": 1.0686,
      "step": 12710
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.7224495989768956e-05,
      "loss": 1.0751,
      "step": 12720
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.7188651911624995e-05,
      "loss": 1.0881,
      "step": 12730
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.7152803298995045e-05,
      "loss": 1.076,
      "step": 12740
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.7116950226150863e-05,
      "loss": 1.1037,
      "step": 12750
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.7081092767373455e-05,
      "loss": 1.1052,
      "step": 12760
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.704523099695291e-05,
      "loss": 1.0942,
      "step": 12770
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.700936498918824e-05,
      "loss": 1.123,
      "step": 12780
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.6973494818387257e-05,
      "loss": 1.1012,
      "step": 12790
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.6937620558866378e-05,
      "loss": 1.0795,
      "step": 12800
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.6901742284950508e-05,
      "loss": 1.1193,
      "step": 12810
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.6865860070972854e-05,
      "loss": 1.1014,
      "step": 12820
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.68299739912748e-05,
      "loss": 1.1011,
      "step": 12830
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.679408412020572e-05,
      "loss": 1.1107,
      "step": 12840
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.675819053212287e-05,
      "loss": 1.076,
      "step": 12850
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.6722293301391183e-05,
      "loss": 1.103,
      "step": 12860
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.6686392502383146e-05,
      "loss": 1.0729,
      "step": 12870
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.6650488209478646e-05,
      "loss": 1.0967,
      "step": 12880
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.6614580497064795e-05,
      "loss": 1.087,
      "step": 12890
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.6578669439535814e-05,
      "loss": 1.0684,
      "step": 12900
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.6542755111292827e-05,
      "loss": 1.0746,
      "step": 12910
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.6506837586743744e-05,
      "loss": 1.1133,
      "step": 12920
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.6470916940303104e-05,
      "loss": 1.1435,
      "step": 12930
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.6434993246391904e-05,
      "loss": 1.1051,
      "step": 12940
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.639906657943747e-05,
      "loss": 1.0866,
      "step": 12950
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.636313701387326e-05,
      "loss": 1.1402,
      "step": 12960
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.6327204624138774e-05,
      "loss": 1.088,
      "step": 12970
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.6291269484679332e-05,
      "loss": 1.0926,
      "step": 12980
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.6255331669945965e-05,
      "loss": 1.0405,
      "step": 12990
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.6219391254395243e-05,
      "loss": 1.1188,
      "step": 13000
    },
    {
      "epoch": 1.61,
      "eval_loss": 1.006820797920227,
      "eval_runtime": 5.7658,
      "eval_samples_per_second": 89.841,
      "eval_steps_per_second": 11.273,
      "step": 13000
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.6183448312489135e-05,
      "loss": 1.0651,
      "step": 13010
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.6147502918694823e-05,
      "loss": 1.0819,
      "step": 13020
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.6111555147484583e-05,
      "loss": 1.0919,
      "step": 13030
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.607560507333562e-05,
      "loss": 1.0267,
      "step": 13040
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.60396527707299e-05,
      "loss": 1.0814,
      "step": 13050
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.6003698314154008e-05,
      "loss": 1.0672,
      "step": 13060
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.5967741778099015e-05,
      "loss": 1.0989,
      "step": 13070
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.5931783237060254e-05,
      "loss": 1.1135,
      "step": 13080
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.589582276553724e-05,
      "loss": 1.1294,
      "step": 13090
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.58598604380335e-05,
      "loss": 1.0852,
      "step": 13100
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.5823896329056384e-05,
      "loss": 1.0922,
      "step": 13110
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.5787930513116943e-05,
      "loss": 1.1142,
      "step": 13120
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.5751963064729752e-05,
      "loss": 1.0643,
      "step": 13130
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.571599405841279e-05,
      "loss": 1.0794,
      "step": 13140
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.5680023568687246e-05,
      "loss": 1.0957,
      "step": 13150
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.5644051670077383e-05,
      "loss": 1.0755,
      "step": 13160
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.56080784371104e-05,
      "loss": 1.0985,
      "step": 13170
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.557210394431624e-05,
      "loss": 1.103,
      "step": 13180
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.5536128266227467e-05,
      "loss": 1.1028,
      "step": 13190
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.55001514773791e-05,
      "loss": 1.097,
      "step": 13200
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.5464173652308447e-05,
      "loss": 1.051,
      "step": 13210
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.5428194865554995e-05,
      "loss": 1.0596,
      "step": 13220
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.5392215191660173e-05,
      "loss": 1.0959,
      "step": 13230
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.5356234705167308e-05,
      "loss": 1.1071,
      "step": 13240
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.5320253480621365e-05,
      "loss": 1.0861,
      "step": 13250
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.528427159256885e-05,
      "loss": 1.1266,
      "step": 13260
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.5248289115557646e-05,
      "loss": 1.06,
      "step": 13270
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.5212306124136858e-05,
      "loss": 1.0775,
      "step": 13280
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.5176322692856662e-05,
      "loss": 1.0725,
      "step": 13290
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.5140338896268134e-05,
      "loss": 1.0847,
      "step": 13300
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.510435480892311e-05,
      "loss": 1.0709,
      "step": 13310
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.5068370505374022e-05,
      "loss": 1.0855,
      "step": 13320
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.5032386060173772e-05,
      "loss": 1.0729,
      "step": 13330
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.499640154787554e-05,
      "loss": 1.1062,
      "step": 13340
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.4960417043032634e-05,
      "loss": 1.0873,
      "step": 13350
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.4924432620198377e-05,
      "loss": 1.078,
      "step": 13360
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.4888448353925885e-05,
      "loss": 1.1389,
      "step": 13370
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.4852464318767976e-05,
      "loss": 1.1065,
      "step": 13380
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.481648058927699e-05,
      "loss": 1.1067,
      "step": 13390
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.478049724000461e-05,
      "loss": 1.1283,
      "step": 13400
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.474451434550177e-05,
      "loss": 1.0986,
      "step": 13410
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.470853198031841e-05,
      "loss": 1.0822,
      "step": 13420
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.4672550219003413e-05,
      "loss": 1.0844,
      "step": 13430
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.4636569136104414e-05,
      "loss": 1.1288,
      "step": 13440
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.460058880616761e-05,
      "loss": 1.0828,
      "step": 13450
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.4564609303737666e-05,
      "loss": 1.1175,
      "step": 13460
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.452863070335751e-05,
      "loss": 1.0762,
      "step": 13470
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.4492653079568227e-05,
      "loss": 1.0889,
      "step": 13480
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.4456676506908866e-05,
      "loss": 1.0886,
      "step": 13490
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.442070105991629e-05,
      "loss": 1.1116,
      "step": 13500
    },
    {
      "epoch": 1.67,
      "eval_loss": 1.002138376235962,
      "eval_runtime": 5.7621,
      "eval_samples_per_second": 89.898,
      "eval_steps_per_second": 11.281,
      "step": 13500
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.4384726813125038e-05,
      "loss": 1.0763,
      "step": 13510
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.434875384106717e-05,
      "loss": 1.0851,
      "step": 13520
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.4312782218272086e-05,
      "loss": 1.1085,
      "step": 13530
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.4276812019266415e-05,
      "loss": 1.0665,
      "step": 13540
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.4240843318573812e-05,
      "loss": 1.1093,
      "step": 13550
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.420487619071484e-05,
      "loss": 1.0694,
      "step": 13560
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.4168910710206808e-05,
      "loss": 1.0724,
      "step": 13570
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.41329469515636e-05,
      "loss": 1.0751,
      "step": 13580
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.409698498929553e-05,
      "loss": 1.0785,
      "step": 13590
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.4061024897909223e-05,
      "loss": 1.0908,
      "step": 13600
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.4025066751907378e-05,
      "loss": 1.0931,
      "step": 13610
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.398911062578871e-05,
      "loss": 1.0761,
      "step": 13620
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.3953156594047715e-05,
      "loss": 1.1235,
      "step": 13630
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.391720473117456e-05,
      "loss": 1.0935,
      "step": 13640
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.388125511165494e-05,
      "loss": 1.0916,
      "step": 13650
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.384530780996986e-05,
      "loss": 1.1348,
      "step": 13660
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.3809362900595573e-05,
      "loss": 1.0946,
      "step": 13670
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.377342045800333e-05,
      "loss": 1.0831,
      "step": 13680
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.3737480556659285e-05,
      "loss": 1.0529,
      "step": 13690
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.370154327102436e-05,
      "loss": 1.1044,
      "step": 13700
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.3665608675554005e-05,
      "loss": 1.0843,
      "step": 13710
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.3629676844698127e-05,
      "loss": 1.0959,
      "step": 13720
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.3593747852900913e-05,
      "loss": 1.0895,
      "step": 13730
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.3557821774600633e-05,
      "loss": 1.1204,
      "step": 13740
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.352189868422957e-05,
      "loss": 1.0996,
      "step": 13750
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.3485978656213763e-05,
      "loss": 1.1129,
      "step": 13760
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.3450061764972942e-05,
      "loss": 1.0746,
      "step": 13770
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.341414808492034e-05,
      "loss": 1.0801,
      "step": 13780
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.3378237690462512e-05,
      "loss": 1.0826,
      "step": 13790
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.3342330655999222e-05,
      "loss": 1.084,
      "step": 13800
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.3306427055923275e-05,
      "loss": 1.0969,
      "step": 13810
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.327052696462035e-05,
      "loss": 1.032,
      "step": 13820
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.3234630456468872e-05,
      "loss": 1.09,
      "step": 13830
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.3198737605839824e-05,
      "loss": 1.0605,
      "step": 13840
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.3162848487096616e-05,
      "loss": 1.064,
      "step": 13850
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.3126963174594947e-05,
      "loss": 1.0752,
      "step": 13860
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.3091081742682596e-05,
      "loss": 1.118,
      "step": 13870
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.3055204265699338e-05,
      "loss": 1.0587,
      "step": 13880
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.3019330817976716e-05,
      "loss": 1.0904,
      "step": 13890
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.298346147383795e-05,
      "loss": 1.0617,
      "step": 13900
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.2947596307597764e-05,
      "loss": 1.083,
      "step": 13910
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.2911735393562206e-05,
      "loss": 1.0648,
      "step": 13920
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.2875878806028518e-05,
      "loss": 1.0828,
      "step": 13930
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.2840026619285005e-05,
      "loss": 1.0687,
      "step": 13940
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.2804178907610803e-05,
      "loss": 1.0991,
      "step": 13950
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.2768335745275836e-05,
      "loss": 1.0864,
      "step": 13960
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.273249720654055e-05,
      "loss": 1.0826,
      "step": 13970
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.2696663365655837e-05,
      "loss": 1.0779,
      "step": 13980
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.2660834296862874e-05,
      "loss": 1.0897,
      "step": 13990
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.2625010074392907e-05,
      "loss": 1.1008,
      "step": 14000
    },
    {
      "epoch": 1.73,
      "eval_loss": 1.0019787549972534,
      "eval_runtime": 5.7595,
      "eval_samples_per_second": 89.939,
      "eval_steps_per_second": 11.286,
      "step": 14000
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.2589190772467183e-05,
      "loss": 1.0905,
      "step": 14010
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.2553376465296727e-05,
      "loss": 1.0939,
      "step": 14020
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.2517567227082217e-05,
      "loss": 1.0887,
      "step": 14030
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.248176313201386e-05,
      "loss": 1.0822,
      "step": 14040
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.244596425427117e-05,
      "loss": 1.0675,
      "step": 14050
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.2410170668022863e-05,
      "loss": 1.0688,
      "step": 14060
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.2374382447426707e-05,
      "loss": 1.06,
      "step": 14070
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.2338599666629333e-05,
      "loss": 1.0736,
      "step": 14080
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.230282239976612e-05,
      "loss": 1.1037,
      "step": 14090
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.2267050720961003e-05,
      "loss": 1.0497,
      "step": 14100
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.2231284704326345e-05,
      "loss": 1.1091,
      "step": 14110
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.219552442396281e-05,
      "loss": 1.113,
      "step": 14120
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.215976995395912e-05,
      "loss": 1.0675,
      "step": 14130
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.212402136839201e-05,
      "loss": 1.1071,
      "step": 14140
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.2088278741325997e-05,
      "loss": 1.0858,
      "step": 14150
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.205254214681325e-05,
      "loss": 1.099,
      "step": 14160
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.201681165889347e-05,
      "loss": 1.0855,
      "step": 14170
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.1981087351593673e-05,
      "loss": 1.0687,
      "step": 14180
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.194536929892807e-05,
      "loss": 1.0719,
      "step": 14190
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.190965757489795e-05,
      "loss": 1.0843,
      "step": 14200
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.1873952253491445e-05,
      "loss": 1.0915,
      "step": 14210
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.1838253408683462e-05,
      "loss": 1.0946,
      "step": 14220
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.1802561114435456e-05,
      "loss": 1.1188,
      "step": 14230
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.176687544469532e-05,
      "loss": 1.116,
      "step": 14240
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.173119647339725e-05,
      "loss": 1.0368,
      "step": 14250
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.169552427446152e-05,
      "loss": 1.1216,
      "step": 14260
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.1659858921794407e-05,
      "loss": 1.0735,
      "step": 14270
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.162420048928798e-05,
      "loss": 1.1231,
      "step": 14280
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.158854905081999e-05,
      "loss": 1.0756,
      "step": 14290
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.1552904680253694e-05,
      "loss": 1.0849,
      "step": 14300
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.1517267451437686e-05,
      "loss": 1.052,
      "step": 14310
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.1481637438205782e-05,
      "loss": 1.0669,
      "step": 14320
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.144601471437686e-05,
      "loss": 1.0848,
      "step": 14330
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.141039935375466e-05,
      "loss": 1.0665,
      "step": 14340
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.1374791430127708e-05,
      "loss": 1.1204,
      "step": 14350
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.1339191017269087e-05,
      "loss": 1.0985,
      "step": 14360
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.130359818893633e-05,
      "loss": 1.113,
      "step": 14370
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.126801301887128e-05,
      "loss": 1.0699,
      "step": 14380
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.1232435580799876e-05,
      "loss": 1.0659,
      "step": 14390
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.1196865948432052e-05,
      "loss": 1.1243,
      "step": 14400
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.116130419546159e-05,
      "loss": 1.0576,
      "step": 14410
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.1125750395565915e-05,
      "loss": 1.0976,
      "step": 14420
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.1090204622406008e-05,
      "loss": 1.0876,
      "step": 14430
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.1054666949626178e-05,
      "loss": 1.0754,
      "step": 14440
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.101913745085399e-05,
      "loss": 1.0733,
      "step": 14450
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.0983616199700063e-05,
      "loss": 1.0759,
      "step": 14460
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.094810326975791e-05,
      "loss": 1.0527,
      "step": 14470
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.0912598734603843e-05,
      "loss": 1.0269,
      "step": 14480
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.0877102667796733e-05,
      "loss": 1.1048,
      "step": 14490
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.084161514287794e-05,
      "loss": 1.0893,
      "step": 14500
    },
    {
      "epoch": 1.79,
      "eval_loss": 1.0002995729446411,
      "eval_runtime": 5.7626,
      "eval_samples_per_second": 89.89,
      "eval_steps_per_second": 11.28,
      "step": 14500
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.0806136233371122e-05,
      "loss": 1.0777,
      "step": 14510
    },
    {
      "epoch": 1.8,
      "learning_rate": 2.0770666012782077e-05,
      "loss": 1.0755,
      "step": 14520
    },
    {
      "epoch": 1.8,
      "learning_rate": 2.0735204554598602e-05,
      "loss": 1.0881,
      "step": 14530
    },
    {
      "epoch": 1.8,
      "learning_rate": 2.0699751932290368e-05,
      "loss": 1.0882,
      "step": 14540
    },
    {
      "epoch": 1.8,
      "learning_rate": 2.0664308219308685e-05,
      "loss": 1.0791,
      "step": 14550
    },
    {
      "epoch": 1.8,
      "learning_rate": 2.062887348908646e-05,
      "loss": 1.1286,
      "step": 14560
    },
    {
      "epoch": 1.8,
      "learning_rate": 2.059344781503796e-05,
      "loss": 1.0334,
      "step": 14570
    },
    {
      "epoch": 1.8,
      "learning_rate": 2.0558031270558678e-05,
      "loss": 1.1084,
      "step": 14580
    },
    {
      "epoch": 1.8,
      "learning_rate": 2.0522623929025237e-05,
      "loss": 1.0731,
      "step": 14590
    },
    {
      "epoch": 1.81,
      "learning_rate": 2.048722586379514e-05,
      "loss": 1.0651,
      "step": 14600
    },
    {
      "epoch": 1.81,
      "learning_rate": 2.0451837148206714e-05,
      "loss": 1.0947,
      "step": 14610
    },
    {
      "epoch": 1.81,
      "learning_rate": 2.0416457855578892e-05,
      "loss": 1.1124,
      "step": 14620
    },
    {
      "epoch": 1.81,
      "learning_rate": 2.038108805921108e-05,
      "loss": 1.0508,
      "step": 14630
    },
    {
      "epoch": 1.81,
      "learning_rate": 2.034572783238304e-05,
      "loss": 1.0964,
      "step": 14640
    },
    {
      "epoch": 1.81,
      "learning_rate": 2.0310377248354666e-05,
      "loss": 1.093,
      "step": 14650
    },
    {
      "epoch": 1.81,
      "learning_rate": 2.02750363803659e-05,
      "loss": 1.0838,
      "step": 14660
    },
    {
      "epoch": 1.81,
      "learning_rate": 2.0239705301636566e-05,
      "loss": 1.0922,
      "step": 14670
    },
    {
      "epoch": 1.82,
      "learning_rate": 2.0204384085366165e-05,
      "loss": 1.084,
      "step": 14680
    },
    {
      "epoch": 1.82,
      "learning_rate": 2.0169072804733812e-05,
      "loss": 1.0713,
      "step": 14690
    },
    {
      "epoch": 1.82,
      "learning_rate": 2.0133771532898e-05,
      "loss": 1.0902,
      "step": 14700
    },
    {
      "epoch": 1.82,
      "learning_rate": 2.0098480342996494e-05,
      "loss": 1.0772,
      "step": 14710
    },
    {
      "epoch": 1.82,
      "learning_rate": 2.0063199308146204e-05,
      "loss": 1.0848,
      "step": 14720
    },
    {
      "epoch": 1.82,
      "learning_rate": 2.0027928501442937e-05,
      "loss": 1.0837,
      "step": 14730
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.999266799596138e-05,
      "loss": 1.0704,
      "step": 14740
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.995741786475483e-05,
      "loss": 1.0567,
      "step": 14750
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.9922178180855094e-05,
      "loss": 1.1145,
      "step": 14760
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.9886949017272366e-05,
      "loss": 1.0776,
      "step": 14770
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.9851730446995004e-05,
      "loss": 1.1325,
      "step": 14780
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.9816522542989437e-05,
      "loss": 1.0749,
      "step": 14790
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.978132537820001e-05,
      "loss": 1.1291,
      "step": 14800
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.974613902554878e-05,
      "loss": 1.086,
      "step": 14810
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.971096355793546e-05,
      "loss": 1.0986,
      "step": 14820
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.9675799048237147e-05,
      "loss": 1.0714,
      "step": 14830
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.9640645569308272e-05,
      "loss": 1.1406,
      "step": 14840
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.9605503193980423e-05,
      "loss": 1.0807,
      "step": 14850
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.9570371995062152e-05,
      "loss": 1.0688,
      "step": 14860
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.9535252045338866e-05,
      "loss": 1.0679,
      "step": 14870
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.9500143417572692e-05,
      "loss": 1.0699,
      "step": 14880
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.9465046184502248e-05,
      "loss": 1.1133,
      "step": 14890
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.9429960418842603e-05,
      "loss": 1.0837,
      "step": 14900
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.9394886193285015e-05,
      "loss": 1.078,
      "step": 14910
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.935982358049687e-05,
      "loss": 1.1155,
      "step": 14920
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.9324772653121483e-05,
      "loss": 1.0921,
      "step": 14930
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.9289733483777944e-05,
      "loss": 1.066,
      "step": 14940
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.9254706145061016e-05,
      "loss": 1.0833,
      "step": 14950
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.9219690709540914e-05,
      "loss": 1.0873,
      "step": 14960
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.918468724976321e-05,
      "loss": 1.0585,
      "step": 14970
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.9149695838248674e-05,
      "loss": 1.0517,
      "step": 14980
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.911471654749309e-05,
      "loss": 1.0816,
      "step": 14990
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.9079749449967148e-05,
      "loss": 1.0257,
      "step": 15000
    },
    {
      "epoch": 1.86,
      "eval_loss": 0.9986115097999573,
      "eval_runtime": 5.7655,
      "eval_samples_per_second": 89.845,
      "eval_steps_per_second": 11.274,
      "step": 15000
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.9044794618116284e-05,
      "loss": 1.0465,
      "step": 15010
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.9009852124360487e-05,
      "loss": 1.0867,
      "step": 15020
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.8974922041094227e-05,
      "loss": 1.0526,
      "step": 15030
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.894000444068622e-05,
      "loss": 1.0998,
      "step": 15040
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.8905099395479348e-05,
      "loss": 1.0879,
      "step": 15050
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.8870206977790485e-05,
      "loss": 1.0923,
      "step": 15060
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.883532725991031e-05,
      "loss": 1.1194,
      "step": 15070
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.8800460314103233e-05,
      "loss": 1.072,
      "step": 15080
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.876560621260717e-05,
      "loss": 1.0629,
      "step": 15090
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.8730765027633428e-05,
      "loss": 1.0837,
      "step": 15100
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.869593683136659e-05,
      "loss": 1.091,
      "step": 15110
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.8661121695964273e-05,
      "loss": 1.0649,
      "step": 15120
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.8626319693557066e-05,
      "loss": 1.0886,
      "step": 15130
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.8591530896248365e-05,
      "loss": 1.0789,
      "step": 15140
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.8556755376114164e-05,
      "loss": 1.1161,
      "step": 15150
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.8521993205203e-05,
      "loss": 1.0907,
      "step": 15160
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.8487244455535696e-05,
      "loss": 1.0725,
      "step": 15170
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.845250919910531e-05,
      "loss": 1.0891,
      "step": 15180
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.8417787507876947e-05,
      "loss": 1.0671,
      "step": 15190
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.8383079453787567e-05,
      "loss": 1.0897,
      "step": 15200
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.8348385108745926e-05,
      "loss": 1.0891,
      "step": 15210
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.8313704544632347e-05,
      "loss": 1.0926,
      "step": 15220
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.82790378332986e-05,
      "loss": 1.0931,
      "step": 15230
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.8244385046567774e-05,
      "loss": 1.0992,
      "step": 15240
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.8209746256234092e-05,
      "loss": 1.0846,
      "step": 15250
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.8175121534062776e-05,
      "loss": 1.0472,
      "step": 15260
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.8140510951789934e-05,
      "loss": 1.0939,
      "step": 15270
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.810591458112233e-05,
      "loss": 1.0917,
      "step": 15280
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.807133249373733e-05,
      "loss": 1.1027,
      "step": 15290
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.803676476128267e-05,
      "loss": 1.1106,
      "step": 15300
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.8002211455376363e-05,
      "loss": 1.0574,
      "step": 15310
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.796767264760655e-05,
      "loss": 1.0676,
      "step": 15320
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.7933148409531286e-05,
      "loss": 1.0656,
      "step": 15330
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.7898638812678496e-05,
      "loss": 1.0868,
      "step": 15340
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.7864143928545745e-05,
      "loss": 1.0756,
      "step": 15350
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.7829663828600096e-05,
      "loss": 1.0795,
      "step": 15360
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.779519858427802e-05,
      "loss": 1.0768,
      "step": 15370
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.776074826698519e-05,
      "loss": 1.0224,
      "step": 15380
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.7726312948096344e-05,
      "loss": 1.0702,
      "step": 15390
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.7691892698955177e-05,
      "loss": 1.0781,
      "step": 15400
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.7657487590874117e-05,
      "loss": 1.0737,
      "step": 15410
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.7623097695134267e-05,
      "loss": 1.1002,
      "step": 15420
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.7588723082985196e-05,
      "loss": 1.0965,
      "step": 15430
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.7554363825644787e-05,
      "loss": 1.1087,
      "step": 15440
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.7520019994299153e-05,
      "loss": 1.0702,
      "step": 15450
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.7485691660102405e-05,
      "loss": 1.0821,
      "step": 15460
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.7451378894176565e-05,
      "loss": 1.0597,
      "step": 15470
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.741708176761142e-05,
      "loss": 1.0824,
      "step": 15480
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.7382800351464316e-05,
      "loss": 1.1146,
      "step": 15490
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.7348534716760084e-05,
      "loss": 1.1132,
      "step": 15500
    },
    {
      "epoch": 1.92,
      "eval_loss": 0.9955187439918518,
      "eval_runtime": 5.7634,
      "eval_samples_per_second": 89.877,
      "eval_steps_per_second": 11.278,
      "step": 15500
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.7314284934490838e-05,
      "loss": 1.0631,
      "step": 15510
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.728005107561585e-05,
      "loss": 1.1108,
      "step": 15520
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.724583321106142e-05,
      "loss": 1.0998,
      "step": 15530
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.7211631411720687e-05,
      "loss": 1.0918,
      "step": 15540
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.717744574845352e-05,
      "loss": 1.0374,
      "step": 15550
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.714327629208636e-05,
      "loss": 1.0973,
      "step": 15560
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.710912311341204e-05,
      "loss": 1.071,
      "step": 15570
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.7074986283189728e-05,
      "loss": 1.0946,
      "step": 15580
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.704086587214466e-05,
      "loss": 1.0998,
      "step": 15590
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.7006761950968082e-05,
      "loss": 1.071,
      "step": 15600
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.6972674590317088e-05,
      "loss": 1.0534,
      "step": 15610
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.693860386081443e-05,
      "loss": 1.1007,
      "step": 15620
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.6904549833048435e-05,
      "loss": 1.077,
      "step": 15630
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.68705125775728e-05,
      "loss": 1.0493,
      "step": 15640
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.6836492164906492e-05,
      "loss": 1.0791,
      "step": 15650
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.6802488665533585e-05,
      "loss": 1.0917,
      "step": 15660
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.6768502149903083e-05,
      "loss": 1.09,
      "step": 15670
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.6734532688428838e-05,
      "loss": 1.1026,
      "step": 15680
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.6700580351489365e-05,
      "loss": 1.0366,
      "step": 15690
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.6666645209427657e-05,
      "loss": 1.1109,
      "step": 15700
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.6632727332551147e-05,
      "loss": 1.0756,
      "step": 15710
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.6598826791131438e-05,
      "loss": 1.0642,
      "step": 15720
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.6564943655404258e-05,
      "loss": 1.0914,
      "step": 15730
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.6531077995569267e-05,
      "loss": 1.0902,
      "step": 15740
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.6497229881789892e-05,
      "loss": 1.0591,
      "step": 15750
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.646339938419324e-05,
      "loss": 1.0667,
      "step": 15760
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.642958657286991e-05,
      "loss": 1.1131,
      "step": 15770
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.6395791517873842e-05,
      "loss": 1.086,
      "step": 15780
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.6362014289222217e-05,
      "loss": 1.0942,
      "step": 15790
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.6328254956895253e-05,
      "loss": 1.1188,
      "step": 15800
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.6294513590836116e-05,
      "loss": 1.0417,
      "step": 15810
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.6260790260950737e-05,
      "loss": 1.075,
      "step": 15820
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.622708503710767e-05,
      "loss": 1.0743,
      "step": 15830
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.6193397989137983e-05,
      "loss": 1.0884,
      "step": 15840
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.615972918683506e-05,
      "loss": 1.0679,
      "step": 15850
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.6126078699954486e-05,
      "loss": 1.0816,
      "step": 15860
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.609244659821393e-05,
      "loss": 1.0676,
      "step": 15870
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.605883295129293e-05,
      "loss": 1.0918,
      "step": 15880
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.6025237828832816e-05,
      "loss": 1.109,
      "step": 15890
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.5991661300436533e-05,
      "loss": 1.063,
      "step": 15900
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.5958103435668486e-05,
      "loss": 1.0995,
      "step": 15910
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.592456430405444e-05,
      "loss": 1.0594,
      "step": 15920
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.589104397508133e-05,
      "loss": 1.1224,
      "step": 15930
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.585754251819713e-05,
      "loss": 1.0781,
      "step": 15940
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.5824060002810744e-05,
      "loss": 1.0535,
      "step": 15950
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.579059649829179e-05,
      "loss": 1.0908,
      "step": 15960
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.5757152073970545e-05,
      "loss": 1.0995,
      "step": 15970
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.5723726799137714e-05,
      "loss": 1.0563,
      "step": 15980
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.5690320743044345e-05,
      "loss": 1.0789,
      "step": 15990
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.5656933974901688e-05,
      "loss": 1.048,
      "step": 16000
    },
    {
      "epoch": 1.98,
      "eval_loss": 0.9951100945472717,
      "eval_runtime": 5.7671,
      "eval_samples_per_second": 89.819,
      "eval_steps_per_second": 11.271,
      "step": 16000
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.562356656388099e-05,
      "loss": 1.0634,
      "step": 16010
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.559021857911344e-05,
      "loss": 1.0741,
      "step": 16020
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.555689008968994e-05,
      "loss": 1.0496,
      "step": 16030
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.5523581164661016e-05,
      "loss": 1.0502,
      "step": 16040
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.5490291873036678e-05,
      "loss": 1.0485,
      "step": 16050
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.545702228378622e-05,
      "loss": 1.0796,
      "step": 16060
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.542377246583815e-05,
      "loss": 1.0784,
      "step": 16070
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.5390542488080008e-05,
      "loss": 1.0884,
      "step": 16080
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.535733241935821e-05,
      "loss": 1.0825,
      "step": 16090
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.532414232847795e-05,
      "loss": 1.1553,
      "step": 16100
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.529097228420302e-05,
      "loss": 1.0951,
      "step": 16110
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.5257822355255657e-05,
      "loss": 1.0855,
      "step": 16120
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.5224692610316471e-05,
      "loss": 1.0998,
      "step": 16130
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.51915831180242e-05,
      "loss": 1.0826,
      "step": 16140
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.5158493946975672e-05,
      "loss": 1.0591,
      "step": 16150
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.5125425165725585e-05,
      "loss": 1.0561,
      "step": 16160
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.5092376842786386e-05,
      "loss": 1.0665,
      "step": 16170
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.5059349046628163e-05,
      "loss": 1.0533,
      "step": 16180
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.5026341845678454e-05,
      "loss": 1.0734,
      "step": 16190
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.4993355308322135e-05,
      "loss": 1.0379,
      "step": 16200
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.4960389502901284e-05,
      "loss": 1.0452,
      "step": 16210
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.492744449771499e-05,
      "loss": 1.0353,
      "step": 16220
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.489452036101929e-05,
      "loss": 1.0286,
      "step": 16230
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.4861617161026967e-05,
      "loss": 1.0727,
      "step": 16240
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.482873496590741e-05,
      "loss": 1.0677,
      "step": 16250
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.4795873843786524e-05,
      "loss": 1.0828,
      "step": 16260
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.4763033862746519e-05,
      "loss": 1.1342,
      "step": 16270
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.4730215090825838e-05,
      "loss": 1.0501,
      "step": 16280
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.469741759601897e-05,
      "loss": 1.056,
      "step": 16290
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.4664641446276295e-05,
      "loss": 1.0772,
      "step": 16300
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.4631886709504019e-05,
      "loss": 1.0691,
      "step": 16310
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.4599153453563943e-05,
      "loss": 1.0591,
      "step": 16320
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.4566441746273376e-05,
      "loss": 1.0511,
      "step": 16330
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.4533751655405015e-05,
      "loss": 1.053,
      "step": 16340
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.450108324868671e-05,
      "loss": 1.0393,
      "step": 16350
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.4468436593801416e-05,
      "loss": 1.0728,
      "step": 16360
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.443581175838706e-05,
      "loss": 1.0796,
      "step": 16370
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.4403208810036295e-05,
      "loss": 1.0782,
      "step": 16380
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.4370627816296468e-05,
      "loss": 1.0577,
      "step": 16390
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.4338068844669434e-05,
      "loss": 1.094,
      "step": 16400
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.4305531962611406e-05,
      "loss": 1.072,
      "step": 16410
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.4273017237532882e-05,
      "loss": 1.0528,
      "step": 16420
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.4240524736798383e-05,
      "loss": 1.0699,
      "step": 16430
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.4208054527726422e-05,
      "loss": 1.0727,
      "step": 16440
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.4175606677589363e-05,
      "loss": 1.0395,
      "step": 16450
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.4143181253613169e-05,
      "loss": 1.0805,
      "step": 16460
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.41107783229774e-05,
      "loss": 1.0824,
      "step": 16470
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.4078397952814993e-05,
      "loss": 1.0447,
      "step": 16480
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.404604021021213e-05,
      "loss": 1.0873,
      "step": 16490
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.4013705162208163e-05,
      "loss": 1.0603,
      "step": 16500
    },
    {
      "epoch": 2.04,
      "eval_loss": 0.9945240616798401,
      "eval_runtime": 5.7637,
      "eval_samples_per_second": 89.872,
      "eval_steps_per_second": 11.277,
      "step": 16500
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.3981392875795363e-05,
      "loss": 1.0761,
      "step": 16510
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.3949103417918873e-05,
      "loss": 1.0865,
      "step": 16520
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.3916836855476545e-05,
      "loss": 1.0787,
      "step": 16530
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.3884593255318778e-05,
      "loss": 1.0605,
      "step": 16540
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.3852372684248429e-05,
      "loss": 1.0953,
      "step": 16550
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.38201752090206e-05,
      "loss": 1.0849,
      "step": 16560
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.378800089634256e-05,
      "loss": 1.0835,
      "step": 16570
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.3755849812873622e-05,
      "loss": 1.084,
      "step": 16580
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.3723722025224916e-05,
      "loss": 1.0312,
      "step": 16590
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.3691617599959345e-05,
      "loss": 1.0514,
      "step": 16600
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.36595366035914e-05,
      "loss": 1.0556,
      "step": 16610
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.3627479102587019e-05,
      "loss": 1.0867,
      "step": 16620
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.3595445163363502e-05,
      "loss": 1.0725,
      "step": 16630
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.356343485228928e-05,
      "loss": 1.062,
      "step": 16640
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.3531448235683867e-05,
      "loss": 1.071,
      "step": 16650
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.3499485379817667e-05,
      "loss": 1.0432,
      "step": 16660
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.3467546350911864e-05,
      "loss": 1.0427,
      "step": 16670
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.3435631215138305e-05,
      "loss": 1.0421,
      "step": 16680
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.3403740038619272e-05,
      "loss": 1.0783,
      "step": 16690
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.337187288742745e-05,
      "loss": 0.9995,
      "step": 16700
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.3340029827585766e-05,
      "loss": 1.0719,
      "step": 16710
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.3308210925067182e-05,
      "loss": 1.0608,
      "step": 16720
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.3276416245794646e-05,
      "loss": 1.0671,
      "step": 16730
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.3244645855640914e-05,
      "loss": 1.041,
      "step": 16740
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.32128998204284e-05,
      "loss": 1.0578,
      "step": 16750
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.3181178205929115e-05,
      "loss": 1.0516,
      "step": 16760
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.3149481077864401e-05,
      "loss": 1.0757,
      "step": 16770
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.3117808501904902e-05,
      "loss": 1.0355,
      "step": 16780
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.3086160543670429e-05,
      "loss": 1.0932,
      "step": 16790
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.3054537268729711e-05,
      "loss": 1.0553,
      "step": 16800
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.3022938742600432e-05,
      "loss": 1.0808,
      "step": 16810
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.2991365030748914e-05,
      "loss": 1.0914,
      "step": 16820
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.2959816198590108e-05,
      "loss": 1.0852,
      "step": 16830
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.292829231148745e-05,
      "loss": 1.0732,
      "step": 16840
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.2896793434752629e-05,
      "loss": 1.0669,
      "step": 16850
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.2865319633645562e-05,
      "loss": 1.0665,
      "step": 16860
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.2833870973374202e-05,
      "loss": 1.0414,
      "step": 16870
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.2802447519094399e-05,
      "loss": 1.0664,
      "step": 16880
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.2771049335909829e-05,
      "loss": 1.0634,
      "step": 16890
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.2739676488871754e-05,
      "loss": 1.0998,
      "step": 16900
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.2708329042978961e-05,
      "loss": 1.0705,
      "step": 16910
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.2677007063177653e-05,
      "loss": 1.0861,
      "step": 16920
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.26457106143612e-05,
      "loss": 1.0848,
      "step": 16930
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.2614439761370128e-05,
      "loss": 1.0748,
      "step": 16940
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.2583194568991921e-05,
      "loss": 1.0478,
      "step": 16950
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.255197510196088e-05,
      "loss": 1.0528,
      "step": 16960
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.2520781424958056e-05,
      "loss": 1.0654,
      "step": 16970
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.2489613602611008e-05,
      "loss": 1.0721,
      "step": 16980
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.2458471699493765e-05,
      "loss": 1.054,
      "step": 16990
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.2427355780126647e-05,
      "loss": 1.0728,
      "step": 17000
    },
    {
      "epoch": 2.1,
      "eval_loss": 0.9933192133903503,
      "eval_runtime": 5.7602,
      "eval_samples_per_second": 89.928,
      "eval_steps_per_second": 11.284,
      "step": 17000
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.2396265908976134e-05,
      "loss": 1.0699,
      "step": 17010
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.2365202150454775e-05,
      "loss": 1.0771,
      "step": 17020
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.2334164568920956e-05,
      "loss": 1.051,
      "step": 17030
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.2303153228678866e-05,
      "loss": 1.0477,
      "step": 17040
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.2272168193978353e-05,
      "loss": 1.0705,
      "step": 17050
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.2241209529014705e-05,
      "loss": 1.0575,
      "step": 17060
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.2210277297928616e-05,
      "loss": 1.0778,
      "step": 17070
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.2179371564806006e-05,
      "loss": 1.0734,
      "step": 17080
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.2148492393677885e-05,
      "loss": 1.0632,
      "step": 17090
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.211763984852027e-05,
      "loss": 1.0603,
      "step": 17100
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.2086813993253956e-05,
      "loss": 1.0633,
      "step": 17110
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.2056014891744468e-05,
      "loss": 1.0286,
      "step": 17120
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.202524260780194e-05,
      "loss": 1.0413,
      "step": 17130
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.1994497205180864e-05,
      "loss": 1.0344,
      "step": 17140
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.1963778747580126e-05,
      "loss": 1.074,
      "step": 17150
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.1933087298642718e-05,
      "loss": 1.0629,
      "step": 17160
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.19024229219557e-05,
      "loss": 1.0703,
      "step": 17170
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.1871785681050075e-05,
      "loss": 1.0224,
      "step": 17180
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.1841175639400565e-05,
      "loss": 1.0682,
      "step": 17190
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.1810592860425582e-05,
      "loss": 1.0918,
      "step": 17200
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.1780037407487046e-05,
      "loss": 1.0554,
      "step": 17210
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.174950934389025e-05,
      "loss": 1.0611,
      "step": 17220
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.171900873288378e-05,
      "loss": 1.0751,
      "step": 17230
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.168853563765929e-05,
      "loss": 1.0518,
      "step": 17240
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.1658090121351448e-05,
      "loss": 1.1095,
      "step": 17250
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.1627672247037824e-05,
      "loss": 1.0296,
      "step": 17260
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.1597282077738638e-05,
      "loss": 1.0758,
      "step": 17270
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.1566919676416805e-05,
      "loss": 1.0651,
      "step": 17280
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.1536585105977621e-05,
      "loss": 1.0812,
      "step": 17290
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.150627842926877e-05,
      "loss": 1.0893,
      "step": 17300
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.1475999709080158e-05,
      "loss": 1.0944,
      "step": 17310
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.1445749008143722e-05,
      "loss": 1.0615,
      "step": 17320
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.1415526389133387e-05,
      "loss": 1.0396,
      "step": 17330
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.1385331914664887e-05,
      "loss": 1.0421,
      "step": 17340
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.1355165647295631e-05,
      "loss": 1.0531,
      "step": 17350
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.1325027649524633e-05,
      "loss": 1.0936,
      "step": 17360
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.1294917983792272e-05,
      "loss": 1.0577,
      "step": 17370
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.1264836712480267e-05,
      "loss": 1.0624,
      "step": 17380
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.1234783897911528e-05,
      "loss": 1.0872,
      "step": 17390
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.1204759602349951e-05,
      "loss": 1.0954,
      "step": 17400
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.1174763888000411e-05,
      "loss": 1.0784,
      "step": 17410
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.114479681700851e-05,
      "loss": 1.0936,
      "step": 17420
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.1114858451460534e-05,
      "loss": 1.0726,
      "step": 17430
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.1084948853383317e-05,
      "loss": 1.0388,
      "step": 17440
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.1055068084744047e-05,
      "loss": 1.0738,
      "step": 17450
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.1025216207450201e-05,
      "loss": 1.0499,
      "step": 17460
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.0995393283349433e-05,
      "loss": 1.0618,
      "step": 17470
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.096559937422934e-05,
      "loss": 1.0639,
      "step": 17480
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.0935834541817489e-05,
      "loss": 1.0589,
      "step": 17490
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.090609884778113e-05,
      "loss": 1.0812,
      "step": 17500
    },
    {
      "epoch": 2.16,
      "eval_loss": 0.9919288754463196,
      "eval_runtime": 5.7632,
      "eval_samples_per_second": 89.881,
      "eval_steps_per_second": 11.279,
      "step": 17500
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.0876392353727177e-05,
      "loss": 1.0623,
      "step": 17510
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.0846715121202075e-05,
      "loss": 1.0766,
      "step": 17520
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.0817067211691584e-05,
      "loss": 1.091,
      "step": 17530
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.0787448686620757e-05,
      "loss": 1.0658,
      "step": 17540
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.0757859607353757e-05,
      "loss": 1.0578,
      "step": 17550
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.072830003519372e-05,
      "loss": 1.0772,
      "step": 17560
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.0698770031382702e-05,
      "loss": 1.079,
      "step": 17570
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.0669269657101433e-05,
      "loss": 1.0738,
      "step": 17580
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.0639798973469286e-05,
      "loss": 1.0909,
      "step": 17590
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.0610358041544152e-05,
      "loss": 1.0681,
      "step": 17600
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.0580946922322211e-05,
      "loss": 1.0665,
      "step": 17610
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.0551565676737954e-05,
      "loss": 1.0916,
      "step": 17620
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.0522214365663912e-05,
      "loss": 1.0602,
      "step": 17630
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.0492893049910632e-05,
      "loss": 1.0712,
      "step": 17640
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.0463601790226536e-05,
      "loss": 1.0618,
      "step": 17650
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.0434340647297722e-05,
      "loss": 1.067,
      "step": 17660
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.040510968174793e-05,
      "loss": 1.0851,
      "step": 17670
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.0375908954138366e-05,
      "loss": 1.0737,
      "step": 17680
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.034673852496759e-05,
      "loss": 1.0422,
      "step": 17690
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.0317598454671403e-05,
      "loss": 1.0621,
      "step": 17700
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.0288488803622682e-05,
      "loss": 1.0654,
      "step": 17710
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.0259409632131281e-05,
      "loss": 1.0501,
      "step": 17720
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.0230361000443955e-05,
      "loss": 1.0637,
      "step": 17730
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.020134296874411e-05,
      "loss": 1.0703,
      "step": 17740
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.0172355597151827e-05,
      "loss": 1.0653,
      "step": 17750
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.0143398945723612e-05,
      "loss": 1.0522,
      "step": 17760
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.0114473074452338e-05,
      "loss": 1.0574,
      "step": 17770
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.0085578043267146e-05,
      "loss": 1.054,
      "step": 17780
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.005671391203322e-05,
      "loss": 1.088,
      "step": 17790
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.002788074055175e-05,
      "loss": 1.0259,
      "step": 17800
    },
    {
      "epoch": 2.2,
      "learning_rate": 9.99907858855982e-06,
      "loss": 1.081,
      "step": 17810
    },
    {
      "epoch": 2.2,
      "learning_rate": 9.970307515730182e-06,
      "loss": 1.0542,
      "step": 17820
    },
    {
      "epoch": 2.21,
      "learning_rate": 9.94156758167126e-06,
      "loss": 1.0599,
      "step": 17830
    },
    {
      "epoch": 2.21,
      "learning_rate": 9.912858845926908e-06,
      "loss": 1.0504,
      "step": 17840
    },
    {
      "epoch": 2.21,
      "learning_rate": 9.884181367976375e-06,
      "loss": 1.0625,
      "step": 17850
    },
    {
      "epoch": 2.21,
      "learning_rate": 9.855535207234164e-06,
      "loss": 1.0617,
      "step": 17860
    },
    {
      "epoch": 2.21,
      "learning_rate": 9.826920423049845e-06,
      "loss": 1.0646,
      "step": 17870
    },
    {
      "epoch": 2.21,
      "learning_rate": 9.798337074708044e-06,
      "loss": 1.0793,
      "step": 17880
    },
    {
      "epoch": 2.21,
      "learning_rate": 9.769785221428199e-06,
      "loss": 1.0962,
      "step": 17890
    },
    {
      "epoch": 2.21,
      "learning_rate": 9.741264922364521e-06,
      "loss": 1.0559,
      "step": 17900
    },
    {
      "epoch": 2.22,
      "learning_rate": 9.712776236605873e-06,
      "loss": 1.0964,
      "step": 17910
    },
    {
      "epoch": 2.22,
      "learning_rate": 9.68431922317557e-06,
      "loss": 1.0542,
      "step": 17920
    },
    {
      "epoch": 2.22,
      "learning_rate": 9.655893941031332e-06,
      "loss": 1.0214,
      "step": 17930
    },
    {
      "epoch": 2.22,
      "learning_rate": 9.627500449065166e-06,
      "loss": 1.0733,
      "step": 17940
    },
    {
      "epoch": 2.22,
      "learning_rate": 9.599138806103153e-06,
      "loss": 1.0449,
      "step": 17950
    },
    {
      "epoch": 2.22,
      "learning_rate": 9.570809070905456e-06,
      "loss": 1.0836,
      "step": 17960
    },
    {
      "epoch": 2.22,
      "learning_rate": 9.542511302166079e-06,
      "loss": 1.0487,
      "step": 17970
    },
    {
      "epoch": 2.22,
      "learning_rate": 9.51424555851281e-06,
      "loss": 1.0717,
      "step": 17980
    },
    {
      "epoch": 2.23,
      "learning_rate": 9.486011898507125e-06,
      "loss": 1.0881,
      "step": 17990
    },
    {
      "epoch": 2.23,
      "learning_rate": 9.45781038064396e-06,
      "loss": 1.0644,
      "step": 18000
    },
    {
      "epoch": 2.23,
      "eval_loss": 0.9902282357215881,
      "eval_runtime": 5.7605,
      "eval_samples_per_second": 89.922,
      "eval_steps_per_second": 11.284,
      "step": 18000
    },
    {
      "epoch": 2.23,
      "learning_rate": 9.429641063351735e-06,
      "loss": 1.1128,
      "step": 18010
    },
    {
      "epoch": 2.23,
      "learning_rate": 9.401504004992088e-06,
      "loss": 1.0436,
      "step": 18020
    },
    {
      "epoch": 2.23,
      "learning_rate": 9.373399263859848e-06,
      "loss": 1.0805,
      "step": 18030
    },
    {
      "epoch": 2.23,
      "learning_rate": 9.345326898182924e-06,
      "loss": 1.0437,
      "step": 18040
    },
    {
      "epoch": 2.23,
      "learning_rate": 9.317286966122091e-06,
      "loss": 1.0439,
      "step": 18050
    },
    {
      "epoch": 2.23,
      "learning_rate": 9.28927952577095e-06,
      "loss": 1.0882,
      "step": 18060
    },
    {
      "epoch": 2.24,
      "learning_rate": 9.261304635155817e-06,
      "loss": 1.0604,
      "step": 18070
    },
    {
      "epoch": 2.24,
      "learning_rate": 9.233362352235508e-06,
      "loss": 1.0871,
      "step": 18080
    },
    {
      "epoch": 2.24,
      "learning_rate": 9.205452734901354e-06,
      "loss": 1.0469,
      "step": 18090
    },
    {
      "epoch": 2.24,
      "learning_rate": 9.17757584097694e-06,
      "loss": 1.0601,
      "step": 18100
    },
    {
      "epoch": 2.24,
      "learning_rate": 9.149731728218089e-06,
      "loss": 1.0607,
      "step": 18110
    },
    {
      "epoch": 2.24,
      "learning_rate": 9.121920454312733e-06,
      "loss": 1.0605,
      "step": 18120
    },
    {
      "epoch": 2.24,
      "learning_rate": 9.094142076880705e-06,
      "loss": 1.0696,
      "step": 18130
    },
    {
      "epoch": 2.24,
      "learning_rate": 9.066396653473724e-06,
      "loss": 1.0658,
      "step": 18140
    },
    {
      "epoch": 2.25,
      "learning_rate": 9.038684241575248e-06,
      "loss": 1.0878,
      "step": 18150
    },
    {
      "epoch": 2.25,
      "learning_rate": 9.01100489860029e-06,
      "loss": 1.0583,
      "step": 18160
    },
    {
      "epoch": 2.25,
      "learning_rate": 8.983358681895407e-06,
      "loss": 1.0577,
      "step": 18170
    },
    {
      "epoch": 2.25,
      "learning_rate": 8.955745648738482e-06,
      "loss": 1.0637,
      "step": 18180
    },
    {
      "epoch": 2.25,
      "learning_rate": 8.928165856338652e-06,
      "loss": 1.0496,
      "step": 18190
    },
    {
      "epoch": 2.25,
      "learning_rate": 8.900619361836232e-06,
      "loss": 1.0465,
      "step": 18200
    },
    {
      "epoch": 2.25,
      "learning_rate": 8.873106222302477e-06,
      "loss": 1.03,
      "step": 18210
    },
    {
      "epoch": 2.25,
      "learning_rate": 8.845626494739608e-06,
      "loss": 1.0413,
      "step": 18220
    },
    {
      "epoch": 2.26,
      "learning_rate": 8.818180236080561e-06,
      "loss": 1.0235,
      "step": 18230
    },
    {
      "epoch": 2.26,
      "learning_rate": 8.790767503188962e-06,
      "loss": 1.0676,
      "step": 18240
    },
    {
      "epoch": 2.26,
      "learning_rate": 8.763388352858998e-06,
      "loss": 1.0355,
      "step": 18250
    },
    {
      "epoch": 2.26,
      "learning_rate": 8.736042841815228e-06,
      "loss": 1.0808,
      "step": 18260
    },
    {
      "epoch": 2.26,
      "learning_rate": 8.708731026712546e-06,
      "loss": 1.0702,
      "step": 18270
    },
    {
      "epoch": 2.26,
      "learning_rate": 8.681452964136055e-06,
      "loss": 1.0801,
      "step": 18280
    },
    {
      "epoch": 2.26,
      "learning_rate": 8.654208710600872e-06,
      "loss": 1.0613,
      "step": 18290
    },
    {
      "epoch": 2.26,
      "learning_rate": 8.626998322552139e-06,
      "loss": 1.0454,
      "step": 18300
    },
    {
      "epoch": 2.26,
      "learning_rate": 8.599821856364764e-06,
      "loss": 1.0556,
      "step": 18310
    },
    {
      "epoch": 2.27,
      "learning_rate": 8.57267936834342e-06,
      "loss": 1.0851,
      "step": 18320
    },
    {
      "epoch": 2.27,
      "learning_rate": 8.545570914722387e-06,
      "loss": 1.1057,
      "step": 18330
    },
    {
      "epoch": 2.27,
      "learning_rate": 8.51849655166539e-06,
      "loss": 1.0475,
      "step": 18340
    },
    {
      "epoch": 2.27,
      "learning_rate": 8.49145633526558e-06,
      "loss": 1.1336,
      "step": 18350
    },
    {
      "epoch": 2.27,
      "learning_rate": 8.464450321545314e-06,
      "loss": 1.0776,
      "step": 18360
    },
    {
      "epoch": 2.27,
      "learning_rate": 8.437478566456102e-06,
      "loss": 1.0983,
      "step": 18370
    },
    {
      "epoch": 2.27,
      "learning_rate": 8.410541125878512e-06,
      "loss": 1.0606,
      "step": 18380
    },
    {
      "epoch": 2.27,
      "learning_rate": 8.383638055621961e-06,
      "loss": 1.0602,
      "step": 18390
    },
    {
      "epoch": 2.28,
      "learning_rate": 8.356769411424683e-06,
      "loss": 1.0135,
      "step": 18400
    },
    {
      "epoch": 2.28,
      "learning_rate": 8.329935248953616e-06,
      "loss": 1.0526,
      "step": 18410
    },
    {
      "epoch": 2.28,
      "learning_rate": 8.303135623804195e-06,
      "loss": 1.0338,
      "step": 18420
    },
    {
      "epoch": 2.28,
      "learning_rate": 8.27637059150037e-06,
      "loss": 1.0835,
      "step": 18430
    },
    {
      "epoch": 2.28,
      "learning_rate": 8.249640207494367e-06,
      "loss": 1.0497,
      "step": 18440
    },
    {
      "epoch": 2.28,
      "learning_rate": 8.22294452716664e-06,
      "loss": 1.0445,
      "step": 18450
    },
    {
      "epoch": 2.28,
      "learning_rate": 8.196283605825777e-06,
      "loss": 1.1042,
      "step": 18460
    },
    {
      "epoch": 2.28,
      "learning_rate": 8.169657498708295e-06,
      "loss": 1.043,
      "step": 18470
    },
    {
      "epoch": 2.29,
      "learning_rate": 8.143066260978632e-06,
      "loss": 1.0481,
      "step": 18480
    },
    {
      "epoch": 2.29,
      "learning_rate": 8.11650994772897e-06,
      "loss": 1.1235,
      "step": 18490
    },
    {
      "epoch": 2.29,
      "learning_rate": 8.089988613979097e-06,
      "loss": 1.0619,
      "step": 18500
    },
    {
      "epoch": 2.29,
      "eval_loss": 0.9900073409080505,
      "eval_runtime": 5.7651,
      "eval_samples_per_second": 89.851,
      "eval_steps_per_second": 11.275,
      "step": 18500
    },
    {
      "epoch": 2.29,
      "learning_rate": 8.063502314676394e-06,
      "loss": 1.0631,
      "step": 18510
    },
    {
      "epoch": 2.29,
      "learning_rate": 8.0370511046956e-06,
      "loss": 1.07,
      "step": 18520
    },
    {
      "epoch": 2.29,
      "learning_rate": 8.010635038838781e-06,
      "loss": 1.0439,
      "step": 18530
    },
    {
      "epoch": 2.29,
      "learning_rate": 7.984254171835207e-06,
      "loss": 1.0427,
      "step": 18540
    },
    {
      "epoch": 2.29,
      "learning_rate": 7.957908558341177e-06,
      "loss": 1.089,
      "step": 18550
    },
    {
      "epoch": 2.3,
      "learning_rate": 7.931598252940003e-06,
      "loss": 1.0909,
      "step": 18560
    },
    {
      "epoch": 2.3,
      "learning_rate": 7.905323310141805e-06,
      "loss": 1.0531,
      "step": 18570
    },
    {
      "epoch": 2.3,
      "learning_rate": 7.879083784383445e-06,
      "loss": 1.093,
      "step": 18580
    },
    {
      "epoch": 2.3,
      "learning_rate": 7.852879730028446e-06,
      "loss": 1.1315,
      "step": 18590
    },
    {
      "epoch": 2.3,
      "learning_rate": 7.826711201366782e-06,
      "loss": 1.0864,
      "step": 18600
    },
    {
      "epoch": 2.3,
      "learning_rate": 7.800578252614853e-06,
      "loss": 1.0541,
      "step": 18610
    },
    {
      "epoch": 2.3,
      "learning_rate": 7.774480937915371e-06,
      "loss": 1.057,
      "step": 18620
    },
    {
      "epoch": 2.3,
      "learning_rate": 7.748419311337157e-06,
      "loss": 1.0301,
      "step": 18630
    },
    {
      "epoch": 2.31,
      "learning_rate": 7.722393426875166e-06,
      "loss": 1.0221,
      "step": 18640
    },
    {
      "epoch": 2.31,
      "learning_rate": 7.696403338450233e-06,
      "loss": 1.0777,
      "step": 18650
    },
    {
      "epoch": 2.31,
      "learning_rate": 7.670449099909072e-06,
      "loss": 1.1002,
      "step": 18660
    },
    {
      "epoch": 2.31,
      "learning_rate": 7.64453076502413e-06,
      "loss": 1.0384,
      "step": 18670
    },
    {
      "epoch": 2.31,
      "learning_rate": 7.618648387493418e-06,
      "loss": 1.0581,
      "step": 18680
    },
    {
      "epoch": 2.31,
      "learning_rate": 7.5928020209405205e-06,
      "loss": 1.0868,
      "step": 18690
    },
    {
      "epoch": 2.31,
      "learning_rate": 7.566991718914349e-06,
      "loss": 1.0309,
      "step": 18700
    },
    {
      "epoch": 2.31,
      "learning_rate": 7.541217534889122e-06,
      "loss": 1.0713,
      "step": 18710
    },
    {
      "epoch": 2.32,
      "learning_rate": 7.515479522264254e-06,
      "loss": 1.0928,
      "step": 18720
    },
    {
      "epoch": 2.32,
      "learning_rate": 7.489777734364173e-06,
      "loss": 1.0612,
      "step": 18730
    },
    {
      "epoch": 2.32,
      "learning_rate": 7.464112224438274e-06,
      "loss": 1.0831,
      "step": 18740
    },
    {
      "epoch": 2.32,
      "learning_rate": 7.438483045660818e-06,
      "loss": 1.0765,
      "step": 18750
    },
    {
      "epoch": 2.32,
      "learning_rate": 7.412890251130741e-06,
      "loss": 1.0827,
      "step": 18760
    },
    {
      "epoch": 2.32,
      "learning_rate": 7.387333893871657e-06,
      "loss": 1.0807,
      "step": 18770
    },
    {
      "epoch": 2.32,
      "learning_rate": 7.361814026831632e-06,
      "loss": 1.0931,
      "step": 18780
    },
    {
      "epoch": 2.32,
      "learning_rate": 7.3363307028831625e-06,
      "loss": 1.094,
      "step": 18790
    },
    {
      "epoch": 2.33,
      "learning_rate": 7.310883974823052e-06,
      "loss": 1.0258,
      "step": 18800
    },
    {
      "epoch": 2.33,
      "learning_rate": 7.285473895372224e-06,
      "loss": 1.0403,
      "step": 18810
    },
    {
      "epoch": 2.33,
      "learning_rate": 7.260100517175744e-06,
      "loss": 1.051,
      "step": 18820
    },
    {
      "epoch": 2.33,
      "learning_rate": 7.234763892802598e-06,
      "loss": 1.0774,
      "step": 18830
    },
    {
      "epoch": 2.33,
      "learning_rate": 7.2094640747456085e-06,
      "loss": 1.0448,
      "step": 18840
    },
    {
      "epoch": 2.33,
      "learning_rate": 7.184201115421402e-06,
      "loss": 1.0844,
      "step": 18850
    },
    {
      "epoch": 2.33,
      "learning_rate": 7.158975067170179e-06,
      "loss": 1.0424,
      "step": 18860
    },
    {
      "epoch": 2.33,
      "learning_rate": 7.133785982255689e-06,
      "loss": 1.0793,
      "step": 18870
    },
    {
      "epoch": 2.34,
      "learning_rate": 7.108633912865128e-06,
      "loss": 1.0719,
      "step": 18880
    },
    {
      "epoch": 2.34,
      "learning_rate": 7.083518911108952e-06,
      "loss": 1.045,
      "step": 18890
    },
    {
      "epoch": 2.34,
      "learning_rate": 7.058441029020871e-06,
      "loss": 1.0393,
      "step": 18900
    },
    {
      "epoch": 2.34,
      "learning_rate": 7.033400318557642e-06,
      "loss": 1.0579,
      "step": 18910
    },
    {
      "epoch": 2.34,
      "learning_rate": 7.0083968315990375e-06,
      "loss": 1.0854,
      "step": 18920
    },
    {
      "epoch": 2.34,
      "learning_rate": 6.98343061994772e-06,
      "loss": 1.0449,
      "step": 18930
    },
    {
      "epoch": 2.34,
      "learning_rate": 6.958501735329081e-06,
      "loss": 1.0826,
      "step": 18940
    },
    {
      "epoch": 2.34,
      "learning_rate": 6.933610229391227e-06,
      "loss": 1.0632,
      "step": 18950
    },
    {
      "epoch": 2.35,
      "learning_rate": 6.908756153704801e-06,
      "loss": 1.0435,
      "step": 18960
    },
    {
      "epoch": 2.35,
      "learning_rate": 6.883939559762873e-06,
      "loss": 1.0785,
      "step": 18970
    },
    {
      "epoch": 2.35,
      "learning_rate": 6.859160498980912e-06,
      "loss": 1.0417,
      "step": 18980
    },
    {
      "epoch": 2.35,
      "learning_rate": 6.834419022696573e-06,
      "loss": 1.0594,
      "step": 18990
    },
    {
      "epoch": 2.35,
      "learning_rate": 6.809715182169663e-06,
      "loss": 1.0554,
      "step": 19000
    },
    {
      "epoch": 2.35,
      "eval_loss": 0.9890897870063782,
      "eval_runtime": 5.7676,
      "eval_samples_per_second": 89.812,
      "eval_steps_per_second": 11.27,
      "step": 19000
    },
    {
      "epoch": 2.35,
      "learning_rate": 6.785049028582041e-06,
      "loss": 1.0474,
      "step": 19010
    },
    {
      "epoch": 2.35,
      "learning_rate": 6.760420613037424e-06,
      "loss": 1.0438,
      "step": 19020
    },
    {
      "epoch": 2.35,
      "learning_rate": 6.735829986561418e-06,
      "loss": 1.0598,
      "step": 19030
    },
    {
      "epoch": 2.36,
      "learning_rate": 6.711277200101271e-06,
      "loss": 1.0698,
      "step": 19040
    },
    {
      "epoch": 2.36,
      "learning_rate": 6.686762304525859e-06,
      "loss": 1.0565,
      "step": 19050
    },
    {
      "epoch": 2.36,
      "learning_rate": 6.662285350625583e-06,
      "loss": 1.0269,
      "step": 19060
    },
    {
      "epoch": 2.36,
      "learning_rate": 6.637846389112176e-06,
      "loss": 1.0512,
      "step": 19070
    },
    {
      "epoch": 2.36,
      "learning_rate": 6.613445470618718e-06,
      "loss": 1.0744,
      "step": 19080
    },
    {
      "epoch": 2.36,
      "learning_rate": 6.589082645699441e-06,
      "loss": 1.0649,
      "step": 19090
    },
    {
      "epoch": 2.36,
      "learning_rate": 6.564757964829632e-06,
      "loss": 1.0165,
      "step": 19100
    },
    {
      "epoch": 2.36,
      "learning_rate": 6.540471478405611e-06,
      "loss": 1.0079,
      "step": 19110
    },
    {
      "epoch": 2.37,
      "learning_rate": 6.516223236744503e-06,
      "loss": 1.0659,
      "step": 19120
    },
    {
      "epoch": 2.37,
      "learning_rate": 6.492013290084231e-06,
      "loss": 1.1184,
      "step": 19130
    },
    {
      "epoch": 2.37,
      "learning_rate": 6.4678416885833895e-06,
      "loss": 1.0566,
      "step": 19140
    },
    {
      "epoch": 2.37,
      "learning_rate": 6.443708482321081e-06,
      "loss": 1.0528,
      "step": 19150
    },
    {
      "epoch": 2.37,
      "learning_rate": 6.41961372129691e-06,
      "loss": 1.0871,
      "step": 19160
    },
    {
      "epoch": 2.37,
      "learning_rate": 6.395557455430814e-06,
      "loss": 1.0871,
      "step": 19170
    },
    {
      "epoch": 2.37,
      "learning_rate": 6.371539734562948e-06,
      "loss": 1.0643,
      "step": 19180
    },
    {
      "epoch": 2.37,
      "learning_rate": 6.347560608453659e-06,
      "loss": 1.0813,
      "step": 19190
    },
    {
      "epoch": 2.38,
      "learning_rate": 6.323620126783284e-06,
      "loss": 1.0611,
      "step": 19200
    },
    {
      "epoch": 2.38,
      "learning_rate": 6.299718339152117e-06,
      "loss": 1.0547,
      "step": 19210
    },
    {
      "epoch": 2.38,
      "learning_rate": 6.275855295080304e-06,
      "loss": 1.0533,
      "step": 19220
    },
    {
      "epoch": 2.38,
      "learning_rate": 6.252031044007681e-06,
      "loss": 1.0614,
      "step": 19230
    },
    {
      "epoch": 2.38,
      "learning_rate": 6.228245635293758e-06,
      "loss": 1.0493,
      "step": 19240
    },
    {
      "epoch": 2.38,
      "learning_rate": 6.204499118217524e-06,
      "loss": 1.08,
      "step": 19250
    },
    {
      "epoch": 2.38,
      "learning_rate": 6.180791541977418e-06,
      "loss": 1.0575,
      "step": 19260
    },
    {
      "epoch": 2.38,
      "learning_rate": 6.157122955691216e-06,
      "loss": 1.1035,
      "step": 19270
    },
    {
      "epoch": 2.38,
      "learning_rate": 6.1334934083958655e-06,
      "loss": 1.0642,
      "step": 19280
    },
    {
      "epoch": 2.39,
      "learning_rate": 6.109902949047483e-06,
      "loss": 1.0261,
      "step": 19290
    },
    {
      "epoch": 2.39,
      "learning_rate": 6.086351626521181e-06,
      "loss": 1.0732,
      "step": 19300
    },
    {
      "epoch": 2.39,
      "learning_rate": 6.062839489610967e-06,
      "loss": 1.068,
      "step": 19310
    },
    {
      "epoch": 2.39,
      "learning_rate": 6.03936658702971e-06,
      "loss": 1.0643,
      "step": 19320
    },
    {
      "epoch": 2.39,
      "learning_rate": 6.015932967408943e-06,
      "loss": 1.0663,
      "step": 19330
    },
    {
      "epoch": 2.39,
      "learning_rate": 5.9925386792988426e-06,
      "loss": 1.0062,
      "step": 19340
    },
    {
      "epoch": 2.39,
      "learning_rate": 5.969183771168102e-06,
      "loss": 1.0926,
      "step": 19350
    },
    {
      "epoch": 2.39,
      "learning_rate": 5.945868291403792e-06,
      "loss": 1.0744,
      "step": 19360
    },
    {
      "epoch": 2.4,
      "learning_rate": 5.922592288311343e-06,
      "loss": 1.0957,
      "step": 19370
    },
    {
      "epoch": 2.4,
      "learning_rate": 5.899355810114349e-06,
      "loss": 1.0783,
      "step": 19380
    },
    {
      "epoch": 2.4,
      "learning_rate": 5.87615890495454e-06,
      "loss": 1.0766,
      "step": 19390
    },
    {
      "epoch": 2.4,
      "learning_rate": 5.853001620891679e-06,
      "loss": 1.0694,
      "step": 19400
    },
    {
      "epoch": 2.4,
      "learning_rate": 5.829884005903385e-06,
      "loss": 1.0366,
      "step": 19410
    },
    {
      "epoch": 2.4,
      "learning_rate": 5.806806107885149e-06,
      "loss": 1.0696,
      "step": 19420
    },
    {
      "epoch": 2.4,
      "learning_rate": 5.783767974650145e-06,
      "loss": 1.0939,
      "step": 19430
    },
    {
      "epoch": 2.4,
      "learning_rate": 5.760769653929149e-06,
      "loss": 1.0562,
      "step": 19440
    },
    {
      "epoch": 2.41,
      "learning_rate": 5.737811193370499e-06,
      "loss": 1.051,
      "step": 19450
    },
    {
      "epoch": 2.41,
      "learning_rate": 5.7148926405398985e-06,
      "loss": 1.0486,
      "step": 19460
    },
    {
      "epoch": 2.41,
      "learning_rate": 5.692014042920391e-06,
      "loss": 1.1251,
      "step": 19470
    },
    {
      "epoch": 2.41,
      "learning_rate": 5.6691754479122615e-06,
      "loss": 1.0897,
      "step": 19480
    },
    {
      "epoch": 2.41,
      "learning_rate": 5.646376902832876e-06,
      "loss": 1.0399,
      "step": 19490
    },
    {
      "epoch": 2.41,
      "learning_rate": 5.6236184549166574e-06,
      "loss": 1.0714,
      "step": 19500
    },
    {
      "epoch": 2.41,
      "eval_loss": 0.9904816746711731,
      "eval_runtime": 5.7632,
      "eval_samples_per_second": 89.881,
      "eval_steps_per_second": 11.279,
      "step": 19500
    },
    {
      "epoch": 2.41,
      "learning_rate": 5.6009001513149474e-06,
      "loss": 1.0641,
      "step": 19510
    },
    {
      "epoch": 2.41,
      "learning_rate": 5.57822203909589e-06,
      "loss": 1.082,
      "step": 19520
    },
    {
      "epoch": 2.42,
      "learning_rate": 5.555584165244407e-06,
      "loss": 1.0902,
      "step": 19530
    },
    {
      "epoch": 2.42,
      "learning_rate": 5.532986576662003e-06,
      "loss": 1.0404,
      "step": 19540
    },
    {
      "epoch": 2.42,
      "learning_rate": 5.51042932016676e-06,
      "loss": 1.0836,
      "step": 19550
    },
    {
      "epoch": 2.42,
      "learning_rate": 5.487912442493187e-06,
      "loss": 1.0784,
      "step": 19560
    },
    {
      "epoch": 2.42,
      "learning_rate": 5.465435990292106e-06,
      "loss": 1.0678,
      "step": 19570
    },
    {
      "epoch": 2.42,
      "learning_rate": 5.443000010130645e-06,
      "loss": 1.0581,
      "step": 19580
    },
    {
      "epoch": 2.42,
      "learning_rate": 5.42060454849202e-06,
      "loss": 1.0626,
      "step": 19590
    },
    {
      "epoch": 2.42,
      "learning_rate": 5.398249651775527e-06,
      "loss": 1.0647,
      "step": 19600
    },
    {
      "epoch": 2.43,
      "learning_rate": 5.3759353662964414e-06,
      "loss": 1.0915,
      "step": 19610
    },
    {
      "epoch": 2.43,
      "learning_rate": 5.353661738285848e-06,
      "loss": 1.0359,
      "step": 19620
    },
    {
      "epoch": 2.43,
      "learning_rate": 5.331428813890649e-06,
      "loss": 1.0776,
      "step": 19630
    },
    {
      "epoch": 2.43,
      "learning_rate": 5.309236639173387e-06,
      "loss": 1.1055,
      "step": 19640
    },
    {
      "epoch": 2.43,
      "learning_rate": 5.2870852601121676e-06,
      "loss": 1.0404,
      "step": 19650
    },
    {
      "epoch": 2.43,
      "learning_rate": 5.264974722600618e-06,
      "loss": 1.051,
      "step": 19660
    },
    {
      "epoch": 2.43,
      "learning_rate": 5.242905072447704e-06,
      "loss": 1.0842,
      "step": 19670
    },
    {
      "epoch": 2.43,
      "learning_rate": 5.220876355377699e-06,
      "loss": 1.0574,
      "step": 19680
    },
    {
      "epoch": 2.44,
      "learning_rate": 5.19888861703009e-06,
      "loss": 1.0616,
      "step": 19690
    },
    {
      "epoch": 2.44,
      "learning_rate": 5.17694190295942e-06,
      "loss": 1.0672,
      "step": 19700
    },
    {
      "epoch": 2.44,
      "learning_rate": 5.155036258635287e-06,
      "loss": 1.0508,
      "step": 19710
    },
    {
      "epoch": 2.44,
      "learning_rate": 5.133171729442154e-06,
      "loss": 1.0613,
      "step": 19720
    },
    {
      "epoch": 2.44,
      "learning_rate": 5.111348360679327e-06,
      "loss": 1.0746,
      "step": 19730
    },
    {
      "epoch": 2.44,
      "learning_rate": 5.089566197560841e-06,
      "loss": 1.0487,
      "step": 19740
    },
    {
      "epoch": 2.44,
      "learning_rate": 5.06782528521533e-06,
      "loss": 1.0551,
      "step": 19750
    },
    {
      "epoch": 2.44,
      "learning_rate": 5.046125668686003e-06,
      "loss": 1.0872,
      "step": 19760
    },
    {
      "epoch": 2.45,
      "learning_rate": 5.024467392930488e-06,
      "loss": 1.0443,
      "step": 19770
    },
    {
      "epoch": 2.45,
      "learning_rate": 5.00285050282075e-06,
      "loss": 1.0377,
      "step": 19780
    },
    {
      "epoch": 2.45,
      "learning_rate": 4.98127504314305e-06,
      "loss": 1.0697,
      "step": 19790
    },
    {
      "epoch": 2.45,
      "learning_rate": 4.959741058597772e-06,
      "loss": 1.0474,
      "step": 19800
    },
    {
      "epoch": 2.45,
      "learning_rate": 4.938248593799388e-06,
      "loss": 1.0707,
      "step": 19810
    },
    {
      "epoch": 2.45,
      "learning_rate": 4.916797693276373e-06,
      "loss": 1.0609,
      "step": 19820
    },
    {
      "epoch": 2.45,
      "learning_rate": 4.895388401471035e-06,
      "loss": 1.0783,
      "step": 19830
    },
    {
      "epoch": 2.45,
      "learning_rate": 4.874020762739526e-06,
      "loss": 1.0768,
      "step": 19840
    },
    {
      "epoch": 2.46,
      "learning_rate": 4.85269482135168e-06,
      "loss": 1.0362,
      "step": 19850
    },
    {
      "epoch": 2.46,
      "learning_rate": 4.831410621490925e-06,
      "loss": 1.0707,
      "step": 19860
    },
    {
      "epoch": 2.46,
      "learning_rate": 4.8101682072542536e-06,
      "loss": 1.0705,
      "step": 19870
    },
    {
      "epoch": 2.46,
      "learning_rate": 4.788967622652027e-06,
      "loss": 1.0915,
      "step": 19880
    },
    {
      "epoch": 2.46,
      "learning_rate": 4.767808911607999e-06,
      "loss": 1.0849,
      "step": 19890
    },
    {
      "epoch": 2.46,
      "learning_rate": 4.746692117959142e-06,
      "loss": 1.0637,
      "step": 19900
    },
    {
      "epoch": 2.46,
      "learning_rate": 4.725617285455564e-06,
      "loss": 1.053,
      "step": 19910
    },
    {
      "epoch": 2.46,
      "learning_rate": 4.704584457760489e-06,
      "loss": 1.0425,
      "step": 19920
    },
    {
      "epoch": 2.47,
      "learning_rate": 4.683593678450063e-06,
      "loss": 1.0338,
      "step": 19930
    },
    {
      "epoch": 2.47,
      "learning_rate": 4.66264499101334e-06,
      "loss": 1.0477,
      "step": 19940
    },
    {
      "epoch": 2.47,
      "learning_rate": 4.641738438852178e-06,
      "loss": 1.0581,
      "step": 19950
    },
    {
      "epoch": 2.47,
      "learning_rate": 4.620874065281103e-06,
      "loss": 1.092,
      "step": 19960
    },
    {
      "epoch": 2.47,
      "learning_rate": 4.6000519135272955e-06,
      "loss": 1.0486,
      "step": 19970
    },
    {
      "epoch": 2.47,
      "learning_rate": 4.579272026730441e-06,
      "loss": 1.0855,
      "step": 19980
    },
    {
      "epoch": 2.47,
      "learning_rate": 4.558534447942639e-06,
      "loss": 1.0563,
      "step": 19990
    },
    {
      "epoch": 2.47,
      "learning_rate": 4.537839220128384e-06,
      "loss": 1.0611,
      "step": 20000
    },
    {
      "epoch": 2.47,
      "eval_loss": 0.9869465231895447,
      "eval_runtime": 5.7657,
      "eval_samples_per_second": 89.841,
      "eval_steps_per_second": 11.274,
      "step": 20000
    },
    {
      "epoch": 2.48,
      "learning_rate": 4.517186386164371e-06,
      "loss": 1.0839,
      "step": 20010
    },
    {
      "epoch": 2.48,
      "learning_rate": 4.4965759888395104e-06,
      "loss": 1.0517,
      "step": 20020
    },
    {
      "epoch": 2.48,
      "learning_rate": 4.476008070854768e-06,
      "loss": 1.0788,
      "step": 20030
    },
    {
      "epoch": 2.48,
      "learning_rate": 4.455482674823089e-06,
      "loss": 1.0623,
      "step": 20040
    },
    {
      "epoch": 2.48,
      "learning_rate": 4.434999843269361e-06,
      "loss": 1.091,
      "step": 20050
    },
    {
      "epoch": 2.48,
      "learning_rate": 4.414559618630237e-06,
      "loss": 1.0405,
      "step": 20060
    },
    {
      "epoch": 2.48,
      "learning_rate": 4.394162043254122e-06,
      "loss": 1.0906,
      "step": 20070
    },
    {
      "epoch": 2.48,
      "learning_rate": 4.373807159401075e-06,
      "loss": 1.0347,
      "step": 20080
    },
    {
      "epoch": 2.49,
      "learning_rate": 4.353495009242667e-06,
      "loss": 1.0644,
      "step": 20090
    },
    {
      "epoch": 2.49,
      "learning_rate": 4.333225634861968e-06,
      "loss": 1.0723,
      "step": 20100
    },
    {
      "epoch": 2.49,
      "learning_rate": 4.312999078253413e-06,
      "loss": 1.0962,
      "step": 20110
    },
    {
      "epoch": 2.49,
      "learning_rate": 4.292815381322701e-06,
      "loss": 1.024,
      "step": 20120
    },
    {
      "epoch": 2.49,
      "learning_rate": 4.272674585886785e-06,
      "loss": 1.067,
      "step": 20130
    },
    {
      "epoch": 2.49,
      "learning_rate": 4.25257673367368e-06,
      "loss": 1.1052,
      "step": 20140
    },
    {
      "epoch": 2.49,
      "learning_rate": 4.23252186632247e-06,
      "loss": 1.0812,
      "step": 20150
    },
    {
      "epoch": 2.49,
      "learning_rate": 4.212510025383173e-06,
      "loss": 1.0749,
      "step": 20160
    },
    {
      "epoch": 2.5,
      "learning_rate": 4.19254125231664e-06,
      "loss": 1.0159,
      "step": 20170
    },
    {
      "epoch": 2.5,
      "learning_rate": 4.172615588494527e-06,
      "loss": 1.0822,
      "step": 20180
    },
    {
      "epoch": 2.5,
      "learning_rate": 4.152733075199161e-06,
      "loss": 1.0417,
      "step": 20190
    },
    {
      "epoch": 2.5,
      "learning_rate": 4.132893753623457e-06,
      "loss": 1.0373,
      "step": 20200
    },
    {
      "epoch": 2.5,
      "learning_rate": 4.1130976648708715e-06,
      "loss": 1.0937,
      "step": 20210
    },
    {
      "epoch": 2.5,
      "learning_rate": 4.093344849955258e-06,
      "loss": 1.0591,
      "step": 20220
    },
    {
      "epoch": 2.5,
      "learning_rate": 4.07363534980085e-06,
      "loss": 1.0878,
      "step": 20230
    },
    {
      "epoch": 2.5,
      "learning_rate": 4.05396920524212e-06,
      "loss": 1.0599,
      "step": 20240
    },
    {
      "epoch": 2.5,
      "learning_rate": 4.034346457023699e-06,
      "loss": 1.1052,
      "step": 20250
    },
    {
      "epoch": 2.51,
      "learning_rate": 4.014767145800355e-06,
      "loss": 1.0553,
      "step": 20260
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.995231312136818e-06,
      "loss": 1.0591,
      "step": 20270
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.975738996507758e-06,
      "loss": 1.0797,
      "step": 20280
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.9562902392976995e-06,
      "loss": 1.1089,
      "step": 20290
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.936885080800884e-06,
      "loss": 1.0423,
      "step": 20300
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.9175235612212615e-06,
      "loss": 1.0509,
      "step": 20310
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.89820572067236e-06,
      "loss": 1.0735,
      "step": 20320
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.878931599177188e-06,
      "loss": 1.048,
      "step": 20330
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.859701236668217e-06,
      "loss": 1.0506,
      "step": 20340
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.840514672987217e-06,
      "loss": 1.0547,
      "step": 20350
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.82137194788525e-06,
      "loss": 1.0245,
      "step": 20360
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.802273101022538e-06,
      "loss": 1.0391,
      "step": 20370
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.783218171968378e-06,
      "loss": 1.0676,
      "step": 20380
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.764207200201114e-06,
      "loss": 1.0483,
      "step": 20390
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.7452402251079847e-06,
      "loss": 1.093,
      "step": 20400
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.726317285985087e-06,
      "loss": 1.1126,
      "step": 20410
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.7074384220373013e-06,
      "loss": 1.0411,
      "step": 20420
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.6886036723781573e-06,
      "loss": 1.0643,
      "step": 20430
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.669813076029824e-06,
      "loss": 1.0714,
      "step": 20440
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.6510666719229758e-06,
      "loss": 1.0847,
      "step": 20450
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.632364498896712e-06,
      "loss": 1.046,
      "step": 20460
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.613706595698532e-06,
      "loss": 1.0629,
      "step": 20470
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.595093000984173e-06,
      "loss": 1.0659,
      "step": 20480
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.576523753317612e-06,
      "loss": 1.0641,
      "step": 20490
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.557998891170933e-06,
      "loss": 1.0941,
      "step": 20500
    },
    {
      "epoch": 2.54,
      "eval_loss": 0.9862278699874878,
      "eval_runtime": 5.7611,
      "eval_samples_per_second": 89.913,
      "eval_steps_per_second": 11.283,
      "step": 20500
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.5395184529242343e-06,
      "loss": 1.0178,
      "step": 20510
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.5210824768656224e-06,
      "loss": 1.0527,
      "step": 20520
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.5026910011910633e-06,
      "loss": 1.0735,
      "step": 20530
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.484344064004308e-06,
      "loss": 1.0969,
      "step": 20540
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.4660417033168718e-06,
      "loss": 1.0721,
      "step": 20550
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.4477839570478785e-06,
      "loss": 1.0474,
      "step": 20560
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.429570863024045e-06,
      "loss": 1.0736,
      "step": 20570
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.4114024589795736e-06,
      "loss": 1.0749,
      "step": 20580
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.3932787825560476e-06,
      "loss": 1.0426,
      "step": 20590
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.3751998713024314e-06,
      "loss": 1.05,
      "step": 20600
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.3571657626748963e-06,
      "loss": 1.0541,
      "step": 20610
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.3391764940368305e-06,
      "loss": 1.0891,
      "step": 20620
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.321232102658703e-06,
      "loss": 1.0593,
      "step": 20630
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.303332625717989e-06,
      "loss": 1.0028,
      "step": 20640
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.285478100299147e-06,
      "loss": 1.0737,
      "step": 20650
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.2676685633934808e-06,
      "loss": 1.0911,
      "step": 20660
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.2499040518990776e-06,
      "loss": 1.0235,
      "step": 20670
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.232184602620772e-06,
      "loss": 1.0846,
      "step": 20680
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.214510252269995e-06,
      "loss": 1.0727,
      "step": 20690
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.19688103746478e-06,
      "loss": 1.0785,
      "step": 20700
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.179296994729636e-06,
      "loss": 1.073,
      "step": 20710
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.1617581604954594e-06,
      "loss": 1.035,
      "step": 20720
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.1442645710995285e-06,
      "loss": 1.078,
      "step": 20730
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.1268162627853315e-06,
      "loss": 1.0674,
      "step": 20740
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.109413271702588e-06,
      "loss": 1.0569,
      "step": 20750
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.0920556339071065e-06,
      "loss": 1.0716,
      "step": 20760
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.0747433853607176e-06,
      "loss": 1.0994,
      "step": 20770
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.0574765619312496e-06,
      "loss": 1.0685,
      "step": 20780
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.0402551993923945e-06,
      "loss": 1.0485,
      "step": 20790
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.0230793334236505e-06,
      "loss": 1.1089,
      "step": 20800
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.0059489996102837e-06,
      "loss": 1.0608,
      "step": 20810
    },
    {
      "epoch": 2.58,
      "learning_rate": 2.9888642334431922e-06,
      "loss": 1.0617,
      "step": 20820
    },
    {
      "epoch": 2.58,
      "learning_rate": 2.9718250703188954e-06,
      "loss": 1.1054,
      "step": 20830
    },
    {
      "epoch": 2.58,
      "learning_rate": 2.9548315455394186e-06,
      "loss": 1.0773,
      "step": 20840
    },
    {
      "epoch": 2.58,
      "learning_rate": 2.9378836943122212e-06,
      "loss": 1.0721,
      "step": 20850
    },
    {
      "epoch": 2.58,
      "learning_rate": 2.9209815517501626e-06,
      "loss": 1.0407,
      "step": 20860
    },
    {
      "epoch": 2.58,
      "learning_rate": 2.9041251528713843e-06,
      "loss": 1.0765,
      "step": 20870
    },
    {
      "epoch": 2.58,
      "learning_rate": 2.8873145325992445e-06,
      "loss": 1.0721,
      "step": 20880
    },
    {
      "epoch": 2.58,
      "learning_rate": 2.870549725762289e-06,
      "loss": 1.0574,
      "step": 20890
    },
    {
      "epoch": 2.59,
      "learning_rate": 2.853830767094112e-06,
      "loss": 1.0529,
      "step": 20900
    },
    {
      "epoch": 2.59,
      "learning_rate": 2.837157691233344e-06,
      "loss": 1.0393,
      "step": 20910
    },
    {
      "epoch": 2.59,
      "learning_rate": 2.8205305327235505e-06,
      "loss": 1.0804,
      "step": 20920
    },
    {
      "epoch": 2.59,
      "learning_rate": 2.8039493260131387e-06,
      "loss": 1.1017,
      "step": 20930
    },
    {
      "epoch": 2.59,
      "learning_rate": 2.7874141054553577e-06,
      "loss": 1.0573,
      "step": 20940
    },
    {
      "epoch": 2.59,
      "learning_rate": 2.7709249053081337e-06,
      "loss": 1.0835,
      "step": 20950
    },
    {
      "epoch": 2.59,
      "learning_rate": 2.7544817597340846e-06,
      "loss": 1.0356,
      "step": 20960
    },
    {
      "epoch": 2.59,
      "learning_rate": 2.7380847028004003e-06,
      "loss": 1.0525,
      "step": 20970
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.7217337684787623e-06,
      "loss": 1.0477,
      "step": 20980
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.7054289906453255e-06,
      "loss": 1.0339,
      "step": 20990
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.68917040308061e-06,
      "loss": 1.0462,
      "step": 21000
    },
    {
      "epoch": 2.6,
      "eval_loss": 0.9858787059783936,
      "eval_runtime": 5.7627,
      "eval_samples_per_second": 89.888,
      "eval_steps_per_second": 11.279,
      "step": 21000
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.6729580394694142e-06,
      "loss": 1.0565,
      "step": 21010
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.656791933400804e-06,
      "loss": 1.0544,
      "step": 21020
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.640672118367976e-06,
      "loss": 1.1011,
      "step": 21030
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.624598627768246e-06,
      "loss": 1.0566,
      "step": 21040
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.6085714949029448e-06,
      "loss": 1.0445,
      "step": 21050
    },
    {
      "epoch": 2.61,
      "learning_rate": 2.592590752977342e-06,
      "loss": 1.0631,
      "step": 21060
    },
    {
      "epoch": 2.61,
      "learning_rate": 2.57665643510063e-06,
      "loss": 1.0176,
      "step": 21070
    },
    {
      "epoch": 2.61,
      "learning_rate": 2.560768574285774e-06,
      "loss": 1.0725,
      "step": 21080
    },
    {
      "epoch": 2.61,
      "learning_rate": 2.544927203449532e-06,
      "loss": 1.09,
      "step": 21090
    },
    {
      "epoch": 2.61,
      "learning_rate": 2.5291323554123198e-06,
      "loss": 1.0854,
      "step": 21100
    },
    {
      "epoch": 2.61,
      "learning_rate": 2.513384062898158e-06,
      "loss": 1.0571,
      "step": 21110
    },
    {
      "epoch": 2.61,
      "learning_rate": 2.4976823585346366e-06,
      "loss": 1.0754,
      "step": 21120
    },
    {
      "epoch": 2.61,
      "learning_rate": 2.4820272748528162e-06,
      "loss": 1.0497,
      "step": 21130
    },
    {
      "epoch": 2.62,
      "learning_rate": 2.4664188442871437e-06,
      "loss": 1.0454,
      "step": 21140
    },
    {
      "epoch": 2.62,
      "learning_rate": 2.450857099175455e-06,
      "loss": 1.0824,
      "step": 21150
    },
    {
      "epoch": 2.62,
      "learning_rate": 2.435342071758809e-06,
      "loss": 1.096,
      "step": 21160
    },
    {
      "epoch": 2.62,
      "learning_rate": 2.4198737941815237e-06,
      "loss": 1.039,
      "step": 21170
    },
    {
      "epoch": 2.62,
      "learning_rate": 2.4044522984910335e-06,
      "loss": 1.0577,
      "step": 21180
    },
    {
      "epoch": 2.62,
      "learning_rate": 2.3890776166378415e-06,
      "loss": 1.0258,
      "step": 21190
    },
    {
      "epoch": 2.62,
      "learning_rate": 2.373749780475487e-06,
      "loss": 1.0604,
      "step": 21200
    },
    {
      "epoch": 2.62,
      "learning_rate": 2.3584688217604317e-06,
      "loss": 1.0978,
      "step": 21210
    },
    {
      "epoch": 2.62,
      "learning_rate": 2.3432347721520282e-06,
      "loss": 1.1037,
      "step": 21220
    },
    {
      "epoch": 2.63,
      "learning_rate": 2.3280476632124403e-06,
      "loss": 1.0763,
      "step": 21230
    },
    {
      "epoch": 2.63,
      "learning_rate": 2.312907526406563e-06,
      "loss": 1.0386,
      "step": 21240
    },
    {
      "epoch": 2.63,
      "learning_rate": 2.2978143931020017e-06,
      "loss": 1.0697,
      "step": 21250
    },
    {
      "epoch": 2.63,
      "learning_rate": 2.2827682945689655e-06,
      "loss": 1.0423,
      "step": 21260
    },
    {
      "epoch": 2.63,
      "learning_rate": 2.2677692619802017e-06,
      "loss": 1.0806,
      "step": 21270
    },
    {
      "epoch": 2.63,
      "learning_rate": 2.2528173264109747e-06,
      "loss": 1.0404,
      "step": 21280
    },
    {
      "epoch": 2.63,
      "learning_rate": 2.2379125188389443e-06,
      "loss": 1.073,
      "step": 21290
    },
    {
      "epoch": 2.63,
      "learning_rate": 2.2230548701441505e-06,
      "loss": 1.0805,
      "step": 21300
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.208244411108923e-06,
      "loss": 1.0579,
      "step": 21310
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.1934811724178076e-06,
      "loss": 1.0522,
      "step": 21320
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.178765184657544e-06,
      "loss": 1.0529,
      "step": 21330
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.164096478316968e-06,
      "loss": 1.0804,
      "step": 21340
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.1494750837869367e-06,
      "loss": 1.0487,
      "step": 21350
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.1349010313603174e-06,
      "loss": 1.0244,
      "step": 21360
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.120374351231866e-06,
      "loss": 1.1087,
      "step": 21370
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.1058950734982123e-06,
      "loss": 1.0515,
      "step": 21380
    },
    {
      "epoch": 2.65,
      "learning_rate": 2.0914632281577694e-06,
      "loss": 1.0904,
      "step": 21390
    },
    {
      "epoch": 2.65,
      "learning_rate": 2.0770788451106663e-06,
      "loss": 1.0499,
      "step": 21400
    },
    {
      "epoch": 2.65,
      "learning_rate": 2.0627419541587286e-06,
      "loss": 1.0872,
      "step": 21410
    },
    {
      "epoch": 2.65,
      "learning_rate": 2.0484525850053505e-06,
      "loss": 1.0333,
      "step": 21420
    },
    {
      "epoch": 2.65,
      "learning_rate": 2.0342107672555013e-06,
      "loss": 1.057,
      "step": 21430
    },
    {
      "epoch": 2.65,
      "learning_rate": 2.0200165304156193e-06,
      "loss": 1.0222,
      "step": 21440
    },
    {
      "epoch": 2.65,
      "learning_rate": 2.0058699038935536e-06,
      "loss": 1.1043,
      "step": 21450
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.9917709169985338e-06,
      "loss": 1.0657,
      "step": 21460
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.97771959894108e-06,
      "loss": 1.0941,
      "step": 21470
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.9637159788329363e-06,
      "loss": 1.0597,
      "step": 21480
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.949760085687058e-06,
      "loss": 1.0796,
      "step": 21490
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.935851948417483e-06,
      "loss": 1.0673,
      "step": 21500
    },
    {
      "epoch": 2.66,
      "eval_loss": 0.9846025109291077,
      "eval_runtime": 5.7645,
      "eval_samples_per_second": 89.861,
      "eval_steps_per_second": 11.276,
      "step": 21500
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.92199159583934e-06,
      "loss": 1.0823,
      "step": 21510
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.908179056668735e-06,
      "loss": 1.0519,
      "step": 21520
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.8944143595227143e-06,
      "loss": 1.0765,
      "step": 21530
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.8806975329192177e-06,
      "loss": 1.0861,
      "step": 21540
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.8670286052769992e-06,
      "loss": 1.064,
      "step": 21550
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.8534076049155663e-06,
      "loss": 1.0632,
      "step": 21560
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.8398345600551481e-06,
      "loss": 1.0269,
      "step": 21570
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.8263094988165918e-06,
      "loss": 1.077,
      "step": 21580
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.8128324492213633e-06,
      "loss": 1.0659,
      "step": 21590
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.799403439191441e-06,
      "loss": 1.0522,
      "step": 21600
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.786022496549261e-06,
      "loss": 1.0669,
      "step": 21610
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.7726896490177003e-06,
      "loss": 1.0454,
      "step": 21620
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.7594049242199678e-06,
      "loss": 1.0545,
      "step": 21630
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.7461683496795861e-06,
      "loss": 1.0455,
      "step": 21640
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.7329799528203183e-06,
      "loss": 1.0398,
      "step": 21650
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.7198397609660939e-06,
      "loss": 1.0348,
      "step": 21660
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.7067478013409966e-06,
      "loss": 1.0592,
      "step": 21670
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.6937041010691657e-06,
      "loss": 1.0702,
      "step": 21680
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.6807086871747591e-06,
      "loss": 1.04,
      "step": 21690
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.6677615865819008e-06,
      "loss": 1.0625,
      "step": 21700
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.6548628261145977e-06,
      "loss": 1.0265,
      "step": 21710
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.6420124324967311e-06,
      "loss": 1.0801,
      "step": 21720
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.6292104323519658e-06,
      "loss": 1.116,
      "step": 21730
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.6164568522036905e-06,
      "loss": 1.0548,
      "step": 21740
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.6037517184749995e-06,
      "loss": 1.0789,
      "step": 21750
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.5910950574885924e-06,
      "loss": 1.0744,
      "step": 21760
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.578486895466763e-06,
      "loss": 1.0454,
      "step": 21770
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.5659272585313161e-06,
      "loss": 1.056,
      "step": 21780
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.5534161727035035e-06,
      "loss": 1.0321,
      "step": 21790
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.5409536639040185e-06,
      "loss": 1.0956,
      "step": 21800
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.528539757952896e-06,
      "loss": 1.0492,
      "step": 21810
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.516174480569471e-06,
      "loss": 1.0647,
      "step": 21820
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.5038578573723428e-06,
      "loss": 1.0394,
      "step": 21830
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.4915899138792822e-06,
      "loss": 1.0763,
      "step": 21840
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.4793706755072385e-06,
      "loss": 1.0714,
      "step": 21850
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.467200167572233e-06,
      "loss": 1.0623,
      "step": 21860
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.4550784152893183e-06,
      "loss": 1.0461,
      "step": 21870
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.4430054437725577e-06,
      "loss": 1.0771,
      "step": 21880
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.4309812780349346e-06,
      "loss": 1.0738,
      "step": 21890
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.4190059429883162e-06,
      "loss": 1.0338,
      "step": 21900
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.4070794634434116e-06,
      "loss": 1.0515,
      "step": 21910
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.3952018641096852e-06,
      "loss": 1.0526,
      "step": 21920
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.3833731695953644e-06,
      "loss": 1.0728,
      "step": 21930
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.371593404407337e-06,
      "loss": 1.0515,
      "step": 21940
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.3598625929511082e-06,
      "loss": 1.1174,
      "step": 21950
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.3481807595307843e-06,
      "loss": 1.0738,
      "step": 21960
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.336547928348972e-06,
      "loss": 1.0821,
      "step": 21970
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.3249641235067777e-06,
      "loss": 1.0564,
      "step": 21980
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.313429369003724e-06,
      "loss": 1.0567,
      "step": 21990
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.3019436887377001e-06,
      "loss": 1.0639,
      "step": 22000
    },
    {
      "epoch": 2.72,
      "eval_loss": 0.9853225946426392,
      "eval_runtime": 5.7651,
      "eval_samples_per_second": 89.85,
      "eval_steps_per_second": 11.275,
      "step": 22000
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.290507106504943e-06,
      "loss": 1.0533,
      "step": 22010
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.2791196459999516e-06,
      "loss": 1.0626,
      "step": 22020
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.2677813308154623e-06,
      "loss": 1.0391,
      "step": 22030
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.256492184442387e-06,
      "loss": 1.0641,
      "step": 22040
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.245252230269761e-06,
      "loss": 1.062,
      "step": 22050
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.234061491584723e-06,
      "loss": 1.0546,
      "step": 22060
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.2229199915724354e-06,
      "loss": 1.0887,
      "step": 22070
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.2118277533160333e-06,
      "loss": 1.0865,
      "step": 22080
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.2007847997966142e-06,
      "loss": 1.0676,
      "step": 22090
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.1897911538931495e-06,
      "loss": 1.0555,
      "step": 22100
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.1788468383824637e-06,
      "loss": 1.051,
      "step": 22110
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.16795187593918e-06,
      "loss": 1.0389,
      "step": 22120
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.157106289135651e-06,
      "loss": 1.0336,
      "step": 22130
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.146310100441958e-06,
      "loss": 1.0335,
      "step": 22140
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.1355633322258253e-06,
      "loss": 1.0916,
      "step": 22150
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.1248660067525868e-06,
      "loss": 1.0762,
      "step": 22160
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.1142181461851475e-06,
      "loss": 1.0352,
      "step": 22170
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.103619772583911e-06,
      "loss": 1.0077,
      "step": 22180
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.093070907906782e-06,
      "loss": 1.0364,
      "step": 22190
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.0825715740090726e-06,
      "loss": 1.0451,
      "step": 22200
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.0721217926434712e-06,
      "loss": 1.0545,
      "step": 22210
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.061721585460021e-06,
      "loss": 1.0817,
      "step": 22220
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.051370974006044e-06,
      "loss": 1.0696,
      "step": 22230
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.0410699797261115e-06,
      "loss": 1.0869,
      "step": 22240
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.0308186239619989e-06,
      "loss": 1.057,
      "step": 22250
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.0206169279526335e-06,
      "loss": 1.1109,
      "step": 22260
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.0104649128340664e-06,
      "loss": 1.0801,
      "step": 22270
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.0003625996394122e-06,
      "loss": 1.053,
      "step": 22280
    },
    {
      "epoch": 2.76,
      "learning_rate": 9.903100092988144e-07,
      "loss": 1.0714,
      "step": 22290
    },
    {
      "epoch": 2.76,
      "learning_rate": 9.803071626394018e-07,
      "loss": 1.0692,
      "step": 22300
    },
    {
      "epoch": 2.76,
      "learning_rate": 9.7035408038523e-07,
      "loss": 1.074,
      "step": 22310
    },
    {
      "epoch": 2.76,
      "learning_rate": 9.604507831572767e-07,
      "loss": 1.0702,
      "step": 22320
    },
    {
      "epoch": 2.76,
      "learning_rate": 9.505972914733568e-07,
      "loss": 1.0599,
      "step": 22330
    },
    {
      "epoch": 2.76,
      "learning_rate": 9.407936257480932e-07,
      "loss": 1.08,
      "step": 22340
    },
    {
      "epoch": 2.76,
      "learning_rate": 9.310398062928938e-07,
      "loss": 1.0631,
      "step": 22350
    },
    {
      "epoch": 2.77,
      "learning_rate": 9.213358533158856e-07,
      "loss": 1.0564,
      "step": 22360
    },
    {
      "epoch": 2.77,
      "learning_rate": 9.116817869218808e-07,
      "loss": 1.0758,
      "step": 22370
    },
    {
      "epoch": 2.77,
      "learning_rate": 9.02077627112341e-07,
      "loss": 1.0887,
      "step": 22380
    },
    {
      "epoch": 2.77,
      "learning_rate": 8.925233937853161e-07,
      "loss": 1.0684,
      "step": 22390
    },
    {
      "epoch": 2.77,
      "learning_rate": 8.830191067354387e-07,
      "loss": 1.0976,
      "step": 22400
    },
    {
      "epoch": 2.77,
      "learning_rate": 8.735647856538492e-07,
      "loss": 1.06,
      "step": 22410
    },
    {
      "epoch": 2.77,
      "learning_rate": 8.641604501281625e-07,
      "loss": 1.0518,
      "step": 22420
    },
    {
      "epoch": 2.77,
      "learning_rate": 8.548061196424512e-07,
      "loss": 1.0146,
      "step": 22430
    },
    {
      "epoch": 2.78,
      "learning_rate": 8.455018135771625e-07,
      "loss": 1.0497,
      "step": 22440
    },
    {
      "epoch": 2.78,
      "learning_rate": 8.362475512091267e-07,
      "loss": 1.0652,
      "step": 22450
    },
    {
      "epoch": 2.78,
      "learning_rate": 8.270433517114761e-07,
      "loss": 1.0454,
      "step": 22460
    },
    {
      "epoch": 2.78,
      "learning_rate": 8.178892341536209e-07,
      "loss": 1.0681,
      "step": 22470
    },
    {
      "epoch": 2.78,
      "learning_rate": 8.087852175012262e-07,
      "loss": 1.0747,
      "step": 22480
    },
    {
      "epoch": 2.78,
      "learning_rate": 7.997313206161427e-07,
      "loss": 1.0458,
      "step": 22490
    },
    {
      "epoch": 2.78,
      "learning_rate": 7.907275622563854e-07,
      "loss": 1.0571,
      "step": 22500
    },
    {
      "epoch": 2.78,
      "eval_loss": 0.9861387610435486,
      "eval_runtime": 5.7641,
      "eval_samples_per_second": 89.866,
      "eval_steps_per_second": 11.277,
      "step": 22500
    },
    {
      "epoch": 2.78,
      "learning_rate": 7.817739610760988e-07,
      "loss": 1.0783,
      "step": 22510
    },
    {
      "epoch": 2.79,
      "learning_rate": 7.728705356254973e-07,
      "loss": 1.081,
      "step": 22520
    },
    {
      "epoch": 2.79,
      "learning_rate": 7.640173043508503e-07,
      "loss": 1.0371,
      "step": 22530
    },
    {
      "epoch": 2.79,
      "learning_rate": 7.552142855944383e-07,
      "loss": 1.0232,
      "step": 22540
    },
    {
      "epoch": 2.79,
      "learning_rate": 7.464614975944945e-07,
      "loss": 1.0483,
      "step": 22550
    },
    {
      "epoch": 2.79,
      "learning_rate": 7.37758958485199e-07,
      "loss": 1.0735,
      "step": 22560
    },
    {
      "epoch": 2.79,
      "learning_rate": 7.291066862966184e-07,
      "loss": 1.0674,
      "step": 22570
    },
    {
      "epoch": 2.79,
      "learning_rate": 7.205046989546771e-07,
      "loss": 1.0794,
      "step": 22580
    },
    {
      "epoch": 2.79,
      "learning_rate": 7.119530142811165e-07,
      "loss": 1.053,
      "step": 22590
    },
    {
      "epoch": 2.8,
      "learning_rate": 7.034516499934584e-07,
      "loss": 1.0833,
      "step": 22600
    },
    {
      "epoch": 2.8,
      "learning_rate": 6.950006237049777e-07,
      "loss": 1.0694,
      "step": 22610
    },
    {
      "epoch": 2.8,
      "learning_rate": 6.865999529246519e-07,
      "loss": 1.0504,
      "step": 22620
    },
    {
      "epoch": 2.8,
      "learning_rate": 6.78249655057131e-07,
      "loss": 1.0902,
      "step": 22630
    },
    {
      "epoch": 2.8,
      "learning_rate": 6.699497474027066e-07,
      "loss": 1.0629,
      "step": 22640
    },
    {
      "epoch": 2.8,
      "learning_rate": 6.617002471572626e-07,
      "loss": 1.0534,
      "step": 22650
    },
    {
      "epoch": 2.8,
      "learning_rate": 6.535011714122546e-07,
      "loss": 1.0808,
      "step": 22660
    },
    {
      "epoch": 2.8,
      "learning_rate": 6.453525371546725e-07,
      "loss": 1.0294,
      "step": 22670
    },
    {
      "epoch": 2.81,
      "learning_rate": 6.37254361266984e-07,
      "loss": 1.0225,
      "step": 22680
    },
    {
      "epoch": 2.81,
      "learning_rate": 6.292066605271319e-07,
      "loss": 1.0803,
      "step": 22690
    },
    {
      "epoch": 2.81,
      "learning_rate": 6.212094516084815e-07,
      "loss": 1.076,
      "step": 22700
    },
    {
      "epoch": 2.81,
      "learning_rate": 6.132627510797795e-07,
      "loss": 1.0371,
      "step": 22710
    },
    {
      "epoch": 2.81,
      "learning_rate": 6.053665754051391e-07,
      "loss": 1.0625,
      "step": 22720
    },
    {
      "epoch": 2.81,
      "learning_rate": 5.975209409439852e-07,
      "loss": 1.0362,
      "step": 22730
    },
    {
      "epoch": 2.81,
      "learning_rate": 5.897258639510406e-07,
      "loss": 1.0622,
      "step": 22740
    },
    {
      "epoch": 2.81,
      "learning_rate": 5.819813605762726e-07,
      "loss": 1.0352,
      "step": 22750
    },
    {
      "epoch": 2.82,
      "learning_rate": 5.742874468648796e-07,
      "loss": 1.0676,
      "step": 22760
    },
    {
      "epoch": 2.82,
      "learning_rate": 5.666441387572413e-07,
      "loss": 1.0404,
      "step": 22770
    },
    {
      "epoch": 2.82,
      "learning_rate": 5.590514520888879e-07,
      "loss": 1.053,
      "step": 22780
    },
    {
      "epoch": 2.82,
      "learning_rate": 5.515094025904777e-07,
      "loss": 1.0321,
      "step": 22790
    },
    {
      "epoch": 2.82,
      "learning_rate": 5.440180058877615e-07,
      "loss": 1.0924,
      "step": 22800
    },
    {
      "epoch": 2.82,
      "learning_rate": 5.36577277501532e-07,
      "loss": 1.1155,
      "step": 22810
    },
    {
      "epoch": 2.82,
      "learning_rate": 5.291872328476222e-07,
      "loss": 1.094,
      "step": 22820
    },
    {
      "epoch": 2.82,
      "learning_rate": 5.218478872368509e-07,
      "loss": 1.0211,
      "step": 22830
    },
    {
      "epoch": 2.83,
      "learning_rate": 5.145592558749968e-07,
      "loss": 1.0557,
      "step": 22840
    },
    {
      "epoch": 2.83,
      "learning_rate": 5.073213538627747e-07,
      "loss": 1.0563,
      "step": 22850
    },
    {
      "epoch": 2.83,
      "learning_rate": 5.001341961957811e-07,
      "loss": 1.0422,
      "step": 22860
    },
    {
      "epoch": 2.83,
      "learning_rate": 4.929977977644995e-07,
      "loss": 1.0748,
      "step": 22870
    },
    {
      "epoch": 2.83,
      "learning_rate": 4.859121733542415e-07,
      "loss": 1.0407,
      "step": 22880
    },
    {
      "epoch": 2.83,
      "learning_rate": 4.788773376451172e-07,
      "loss": 1.0422,
      "step": 22890
    },
    {
      "epoch": 2.83,
      "learning_rate": 4.7189330521202336e-07,
      "loss": 1.0907,
      "step": 22900
    },
    {
      "epoch": 2.83,
      "learning_rate": 4.64960090524591e-07,
      "loss": 1.0633,
      "step": 22910
    },
    {
      "epoch": 2.84,
      "learning_rate": 4.580777079471771e-07,
      "loss": 1.0597,
      "step": 22920
    },
    {
      "epoch": 2.84,
      "learning_rate": 4.5124617173881724e-07,
      "loss": 1.0579,
      "step": 22930
    },
    {
      "epoch": 2.84,
      "learning_rate": 4.44465496053198e-07,
      "loss": 1.0928,
      "step": 22940
    },
    {
      "epoch": 2.84,
      "learning_rate": 4.3773569493864295e-07,
      "loss": 1.0277,
      "step": 22950
    },
    {
      "epoch": 2.84,
      "learning_rate": 4.310567823380657e-07,
      "loss": 1.0687,
      "step": 22960
    },
    {
      "epoch": 2.84,
      "learning_rate": 4.2442877208895004e-07,
      "loss": 1.0341,
      "step": 22970
    },
    {
      "epoch": 2.84,
      "learning_rate": 4.1785167792332546e-07,
      "loss": 1.0519,
      "step": 22980
    },
    {
      "epoch": 2.84,
      "learning_rate": 4.1132551346771406e-07,
      "loss": 1.0239,
      "step": 22990
    },
    {
      "epoch": 2.85,
      "learning_rate": 4.0485029224314465e-07,
      "loss": 1.0654,
      "step": 23000
    },
    {
      "epoch": 2.85,
      "eval_loss": 0.9845592975616455,
      "eval_runtime": 5.7608,
      "eval_samples_per_second": 89.918,
      "eval_steps_per_second": 11.283,
      "step": 23000
    },
    {
      "epoch": 2.85,
      "learning_rate": 3.984260276650886e-07,
      "loss": 1.0744,
      "step": 23010
    },
    {
      "epoch": 2.85,
      "learning_rate": 3.920527330434409e-07,
      "loss": 1.0382,
      "step": 23020
    },
    {
      "epoch": 2.85,
      "learning_rate": 3.8573042158250307e-07,
      "loss": 1.0473,
      "step": 23030
    },
    {
      "epoch": 2.85,
      "learning_rate": 3.7945910638094716e-07,
      "loss": 1.0671,
      "step": 23040
    },
    {
      "epoch": 2.85,
      "learning_rate": 3.7323880043179105e-07,
      "loss": 1.0436,
      "step": 23050
    },
    {
      "epoch": 2.85,
      "learning_rate": 3.6706951662237033e-07,
      "loss": 1.0827,
      "step": 23060
    },
    {
      "epoch": 2.85,
      "learning_rate": 3.609512677343052e-07,
      "loss": 1.0541,
      "step": 23070
    },
    {
      "epoch": 2.85,
      "learning_rate": 3.548840664434949e-07,
      "loss": 1.0506,
      "step": 23080
    },
    {
      "epoch": 2.86,
      "learning_rate": 3.4886792532006497e-07,
      "loss": 1.0357,
      "step": 23090
    },
    {
      "epoch": 2.86,
      "learning_rate": 3.4290285682835875e-07,
      "loss": 1.0679,
      "step": 23100
    },
    {
      "epoch": 2.86,
      "learning_rate": 3.3698887332691e-07,
      "loss": 1.0591,
      "step": 23110
    },
    {
      "epoch": 2.86,
      "learning_rate": 3.311259870684008e-07,
      "loss": 1.0922,
      "step": 23120
    },
    {
      "epoch": 2.86,
      "learning_rate": 3.253142101996648e-07,
      "loss": 1.0554,
      "step": 23130
    },
    {
      "epoch": 2.86,
      "learning_rate": 3.1955355476163964e-07,
      "loss": 1.0365,
      "step": 23140
    },
    {
      "epoch": 2.86,
      "learning_rate": 3.138440326893449e-07,
      "loss": 1.0632,
      "step": 23150
    },
    {
      "epoch": 2.86,
      "learning_rate": 3.0818565581186564e-07,
      "loss": 1.0702,
      "step": 23160
    },
    {
      "epoch": 2.87,
      "learning_rate": 3.0257843585232705e-07,
      "loss": 1.0591,
      "step": 23170
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.970223844278558e-07,
      "loss": 1.0756,
      "step": 23180
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.915175130495773e-07,
      "loss": 1.0669,
      "step": 23190
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.86063833122574e-07,
      "loss": 1.1059,
      "step": 23200
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.8066135594587705e-07,
      "loss": 1.0326,
      "step": 23210
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.753100927124247e-07,
      "loss": 1.0812,
      "step": 23220
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.7001005450905394e-07,
      "loss": 1.0917,
      "step": 23230
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.647612523164811e-07,
      "loss": 1.0918,
      "step": 23240
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.5956369700925187e-07,
      "loss": 1.0586,
      "step": 23250
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.544173993557608e-07,
      "loss": 1.0021,
      "step": 23260
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.4932237001818737e-07,
      "loss": 1.0642,
      "step": 23270
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.4427861955249877e-07,
      "loss": 1.11,
      "step": 23280
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.3928615840842495e-07,
      "loss": 1.0417,
      "step": 23290
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.343449969294309e-07,
      "loss": 1.0623,
      "step": 23300
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.294551453526972e-07,
      "loss": 1.0661,
      "step": 23310
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.2461661380910327e-07,
      "loss": 1.0312,
      "step": 23320
    },
    {
      "epoch": 2.89,
      "learning_rate": 2.1982941232319698e-07,
      "loss": 1.05,
      "step": 23330
    },
    {
      "epoch": 2.89,
      "learning_rate": 2.1509355081318905e-07,
      "loss": 1.0518,
      "step": 23340
    },
    {
      "epoch": 2.89,
      "learning_rate": 2.1040903909091136e-07,
      "loss": 1.0555,
      "step": 23350
    },
    {
      "epoch": 2.89,
      "learning_rate": 2.0577588686181705e-07,
      "loss": 1.0714,
      "step": 23360
    },
    {
      "epoch": 2.89,
      "learning_rate": 2.0119410372494707e-07,
      "loss": 1.06,
      "step": 23370
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.9666369917291927e-07,
      "loss": 1.0408,
      "step": 23380
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.9218468259190324e-07,
      "loss": 1.0791,
      "step": 23390
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.8775706326160104e-07,
      "loss": 1.0628,
      "step": 23400
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.833808503552248e-07,
      "loss": 1.0807,
      "step": 23410
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.7905605293948857e-07,
      "loss": 1.0926,
      "step": 23420
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.747826799745833e-07,
      "loss": 1.0784,
      "step": 23430
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.705607403141546e-07,
      "loss": 1.0559,
      "step": 23440
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.6639024270528603e-07,
      "loss": 1.0738,
      "step": 23450
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.6227119578848537e-07,
      "loss": 1.0601,
      "step": 23460
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.582036080976651e-07,
      "loss": 1.0578,
      "step": 23470
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.5418748806012295e-07,
      "loss": 1.0649,
      "step": 23480
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.5022284399651975e-07,
      "loss": 1.101,
      "step": 23490
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.4630968412087386e-07,
      "loss": 1.0464,
      "step": 23500
    },
    {
      "epoch": 2.91,
      "eval_loss": 0.9841265678405762,
      "eval_runtime": 5.7616,
      "eval_samples_per_second": 89.905,
      "eval_steps_per_second": 11.282,
      "step": 23500
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.4244801654053896e-07,
      "loss": 1.0603,
      "step": 23510
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.3863784925618184e-07,
      "loss": 1.084,
      "step": 23520
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.348791901617713e-07,
      "loss": 1.0628,
      "step": 23530
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.3117204704455877e-07,
      "loss": 1.0322,
      "step": 23540
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.275164275850671e-07,
      "loss": 1.0576,
      "step": 23550
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.2391233935707126e-07,
      "loss": 1.0826,
      "step": 23560
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.2035978982758155e-07,
      "loss": 1.0501,
      "step": 23570
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.1685878635683267e-07,
      "loss": 1.0405,
      "step": 23580
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.1340933619825577e-07,
      "loss": 1.1107,
      "step": 23590
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.1001144649848693e-07,
      "loss": 1.0154,
      "step": 23600
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.0666512429732823e-07,
      "loss": 1.0645,
      "step": 23610
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.0337037652774496e-07,
      "loss": 1.0655,
      "step": 23620
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.0012721001585734e-07,
      "loss": 1.072,
      "step": 23630
    },
    {
      "epoch": 2.92,
      "learning_rate": 9.693563148090723e-08,
      "loss": 1.0466,
      "step": 23640
    },
    {
      "epoch": 2.93,
      "learning_rate": 9.379564753526082e-08,
      "loss": 1.0613,
      "step": 23650
    },
    {
      "epoch": 2.93,
      "learning_rate": 9.070726468439206e-08,
      "loss": 1.0306,
      "step": 23660
    },
    {
      "epoch": 2.93,
      "learning_rate": 8.767048932686318e-08,
      "loss": 1.0568,
      "step": 23670
    },
    {
      "epoch": 2.93,
      "learning_rate": 8.468532775431914e-08,
      "loss": 1.0501,
      "step": 23680
    },
    {
      "epoch": 2.93,
      "learning_rate": 8.17517861514655e-08,
      "loss": 1.0954,
      "step": 23690
    },
    {
      "epoch": 2.93,
      "learning_rate": 7.886987059606277e-08,
      "loss": 1.0636,
      "step": 23700
    },
    {
      "epoch": 2.93,
      "learning_rate": 7.603958705891535e-08,
      "loss": 1.0391,
      "step": 23710
    },
    {
      "epoch": 2.93,
      "learning_rate": 7.326094140384932e-08,
      "loss": 1.0617,
      "step": 23720
    },
    {
      "epoch": 2.94,
      "learning_rate": 7.053393938771246e-08,
      "loss": 1.0403,
      "step": 23730
    },
    {
      "epoch": 2.94,
      "learning_rate": 6.785858666035205e-08,
      "loss": 1.079,
      "step": 23740
    },
    {
      "epoch": 2.94,
      "learning_rate": 6.523488876460648e-08,
      "loss": 1.0704,
      "step": 23750
    },
    {
      "epoch": 2.94,
      "learning_rate": 6.266285113630254e-08,
      "loss": 1.0731,
      "step": 23760
    },
    {
      "epoch": 2.94,
      "learning_rate": 6.014247910423044e-08,
      "loss": 1.0713,
      "step": 23770
    },
    {
      "epoch": 2.94,
      "learning_rate": 5.767377789013539e-08,
      "loss": 1.0578,
      "step": 23780
    },
    {
      "epoch": 2.94,
      "learning_rate": 5.525675260871776e-08,
      "loss": 1.0642,
      "step": 23790
    },
    {
      "epoch": 2.94,
      "learning_rate": 5.289140826760797e-08,
      "loss": 1.0856,
      "step": 23800
    },
    {
      "epoch": 2.95,
      "learning_rate": 5.057774976737206e-08,
      "loss": 1.0541,
      "step": 23810
    },
    {
      "epoch": 2.95,
      "learning_rate": 4.8315781901481225e-08,
      "loss": 1.0546,
      "step": 23820
    },
    {
      "epoch": 2.95,
      "learning_rate": 4.6105509356322854e-08,
      "loss": 1.041,
      "step": 23830
    },
    {
      "epoch": 2.95,
      "learning_rate": 4.394693671117278e-08,
      "loss": 1.0959,
      "step": 23840
    },
    {
      "epoch": 2.95,
      "learning_rate": 4.184006843820087e-08,
      "loss": 1.0648,
      "step": 23850
    },
    {
      "epoch": 2.95,
      "learning_rate": 3.9784908902454345e-08,
      "loss": 1.0374,
      "step": 23860
    },
    {
      "epoch": 2.95,
      "learning_rate": 3.778146236184388e-08,
      "loss": 1.0584,
      "step": 23870
    },
    {
      "epoch": 2.95,
      "learning_rate": 3.5829732967143646e-08,
      "loss": 1.0565,
      "step": 23880
    },
    {
      "epoch": 2.96,
      "learning_rate": 3.3929724761980197e-08,
      "loss": 1.0571,
      "step": 23890
    },
    {
      "epoch": 2.96,
      "learning_rate": 3.2081441682826895e-08,
      "loss": 1.0379,
      "step": 23900
    },
    {
      "epoch": 2.96,
      "learning_rate": 3.0284887558981735e-08,
      "loss": 1.0255,
      "step": 23910
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.8540066112581198e-08,
      "loss": 1.0542,
      "step": 23920
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.6846980958569745e-08,
      "loss": 1.054,
      "step": 23930
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.5205635604713673e-08,
      "loss": 1.066,
      "step": 23940
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.3616033451581698e-08,
      "loss": 1.0623,
      "step": 23950
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.207817779253385e-08,
      "loss": 1.0531,
      "step": 23960
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.0592071813727022e-08,
      "loss": 1.0666,
      "step": 23970
    },
    {
      "epoch": 2.97,
      "learning_rate": 1.9157718594103868e-08,
      "loss": 1.0586,
      "step": 23980
    },
    {
      "epoch": 2.97,
      "learning_rate": 1.7775121105378933e-08,
      "loss": 1.0567,
      "step": 23990
    },
    {
      "epoch": 2.97,
      "learning_rate": 1.6444282212041416e-08,
      "loss": 1.0682,
      "step": 24000
    },
    {
      "epoch": 2.97,
      "eval_loss": 0.9839575290679932,
      "eval_runtime": 5.7619,
      "eval_samples_per_second": 89.901,
      "eval_steps_per_second": 11.281,
      "step": 24000
    }
  ],
  "max_steps": 24252,
  "num_train_epochs": 3,
  "total_flos": 3.86042215804416e+17,
  "trial_name": null,
  "trial_params": null
}