{
  "best_metric": 0.39511793851852417,
  "best_model_checkpoint": "./exper_batch_16_e8/checkpoint-4900",
  "epoch": 8.0,
  "global_step": 5112,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 0.00019960876369327075,
      "loss": 4.9227,
      "step": 10
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0001992175273865415,
      "loss": 4.809,
      "step": 20
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0001988262910798122,
      "loss": 4.6411,
      "step": 30
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019843505477308294,
      "loss": 4.4483,
      "step": 40
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019804381846635368,
      "loss": 4.3374,
      "step": 50
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019765258215962445,
      "loss": 4.2538,
      "step": 60
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019726134585289516,
      "loss": 4.1378,
      "step": 70
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0001968701095461659,
      "loss": 4.0433,
      "step": 80
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019647887323943664,
      "loss": 4.0425,
      "step": 90
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00019608763693270738,
      "loss": 3.8115,
      "step": 100
    },
    {
      "epoch": 0.16,
      "eval_accuracy": 0.18619892058596763,
      "eval_loss": 3.7948496341705322,
      "eval_runtime": 43.0404,
      "eval_samples_per_second": 60.269,
      "eval_steps_per_second": 7.551,
      "step": 100
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0001956964006259781,
      "loss": 3.8373,
      "step": 110
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00019530516431924883,
      "loss": 3.6742,
      "step": 120
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00019491392801251957,
      "loss": 3.581,
      "step": 130
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0001945226917057903,
      "loss": 3.4363,
      "step": 140
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00019413145539906105,
      "loss": 3.433,
      "step": 150
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00019374021909233179,
      "loss": 3.3784,
      "step": 160
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00019334898278560253,
      "loss": 3.2506,
      "step": 170
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00019295774647887326,
      "loss": 3.1149,
      "step": 180
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00019256651017214398,
      "loss": 3.0718,
      "step": 190
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00019217527386541472,
      "loss": 3.1194,
      "step": 200
    },
    {
      "epoch": 0.31,
      "eval_accuracy": 0.32806476484194297,
      "eval_loss": 3.0119543075561523,
      "eval_runtime": 42.8934,
      "eval_samples_per_second": 60.476,
      "eval_steps_per_second": 7.577,
      "step": 200
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00019178403755868546,
      "loss": 2.8329,
      "step": 210
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0001913928012519562,
      "loss": 2.9553,
      "step": 220
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0001910015649452269,
      "loss": 2.9676,
      "step": 230
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00019061032863849765,
      "loss": 2.7578,
      "step": 240
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00019021909233176841,
      "loss": 2.6164,
      "step": 250
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00018982785602503915,
      "loss": 2.7734,
      "step": 260
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00018943661971830987,
      "loss": 2.601,
      "step": 270
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0001890453834115806,
      "loss": 2.8081,
      "step": 280
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00018865414710485134,
      "loss": 2.5308,
      "step": 290
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00018826291079812208,
      "loss": 2.3703,
      "step": 300
    },
    {
      "epoch": 0.47,
      "eval_accuracy": 0.4425597532767926,
      "eval_loss": 2.479144811630249,
      "eval_runtime": 42.6319,
      "eval_samples_per_second": 60.846,
      "eval_steps_per_second": 7.623,
      "step": 300
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0001878716744913928,
      "loss": 2.5485,
      "step": 310
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00018748043818466354,
      "loss": 2.4543,
      "step": 320
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00018708920187793427,
      "loss": 2.3969,
      "step": 330
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00018669796557120501,
      "loss": 2.4463,
      "step": 340
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.00018630672926447575,
      "loss": 2.2321,
      "step": 350
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0001859154929577465,
      "loss": 2.1456,
      "step": 360
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00018552425665101723,
      "loss": 2.152,
      "step": 370
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00018513302034428797,
      "loss": 2.1674,
      "step": 380
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00018474178403755868,
      "loss": 2.1503,
      "step": 390
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00018435054773082942,
      "loss": 2.07,
      "step": 400
    },
    {
      "epoch": 0.63,
      "eval_accuracy": 0.5,
      "eval_loss": 2.171962261199951,
      "eval_runtime": 43.3053,
      "eval_samples_per_second": 59.9,
      "eval_steps_per_second": 7.505,
      "step": 400
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00018395931142410016,
      "loss": 2.1113,
      "step": 410
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0001835680751173709,
      "loss": 2.0248,
      "step": 420
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00018317683881064161,
      "loss": 1.9278,
      "step": 430
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00018278560250391235,
      "loss": 1.8649,
      "step": 440
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00018239436619718312,
      "loss": 1.9462,
      "step": 450
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00018200312989045386,
      "loss": 2.0817,
      "step": 460
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00018161189358372457,
      "loss": 1.8968,
      "step": 470
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0001812206572769953,
      "loss": 1.8535,
      "step": 480
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00018082942097026605,
      "loss": 1.8609,
      "step": 490
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0001804381846635368,
      "loss": 1.6847,
      "step": 500
    },
    {
      "epoch": 0.78,
      "eval_accuracy": 0.5956052428681573,
      "eval_loss": 1.7290887832641602,
      "eval_runtime": 43.1676,
      "eval_samples_per_second": 60.091,
      "eval_steps_per_second": 7.529,
      "step": 500
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0001800469483568075,
      "loss": 1.6108,
      "step": 510
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00017965571205007824,
      "loss": 1.6568,
      "step": 520
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00017926447574334898,
      "loss": 1.6968,
      "step": 530
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00017887323943661972,
      "loss": 1.5102,
      "step": 540
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00017848200312989046,
      "loss": 1.5518,
      "step": 550
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0001780907668231612,
      "loss": 1.6318,
      "step": 560
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00017769953051643194,
      "loss": 1.4528,
      "step": 570
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.00017730829420970268,
      "loss": 1.5259,
      "step": 580
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00017691705790297342,
      "loss": 1.3553,
      "step": 590
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00017652582159624413,
      "loss": 1.3821,
      "step": 600
    },
    {
      "epoch": 0.94,
      "eval_accuracy": 0.6299151888974557,
      "eval_loss": 1.4777374267578125,
      "eval_runtime": 42.6804,
      "eval_samples_per_second": 60.777,
      "eval_steps_per_second": 7.615,
      "step": 600
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00017613458528951487,
      "loss": 1.5701,
      "step": 610
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0001757433489827856,
      "loss": 1.4866,
      "step": 620
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00017535211267605635,
      "loss": 1.4713,
      "step": 630
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0001749608763693271,
      "loss": 1.2937,
      "step": 640
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00017456964006259783,
      "loss": 1.169,
      "step": 650
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00017417840375586857,
      "loss": 1.1436,
      "step": 660
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0001737871674491393,
      "loss": 1.1583,
      "step": 670
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00017339593114241002,
      "loss": 0.9302,
      "step": 680
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00017300469483568076,
      "loss": 1.113,
      "step": 690
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0001726134585289515,
      "loss": 0.9498,
      "step": 700
    },
    {
      "epoch": 1.1,
      "eval_accuracy": 0.6680801850424055,
      "eval_loss": 1.2934883832931519,
      "eval_runtime": 42.6632,
      "eval_samples_per_second": 60.802,
      "eval_steps_per_second": 7.618,
      "step": 700
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00017222222222222224,
      "loss": 0.9322,
      "step": 710
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00017183098591549295,
      "loss": 0.9719,
      "step": 720
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0001714397496087637,
      "loss": 0.9679,
      "step": 730
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00017104851330203443,
      "loss": 0.8539,
      "step": 740
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0001706572769953052,
      "loss": 0.971,
      "step": 750
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0001702660406885759,
      "loss": 0.9654,
      "step": 760
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00016987480438184665,
      "loss": 0.7825,
      "step": 770
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00016948356807511739,
      "loss": 1.0415,
      "step": 780
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00016909233176838813,
      "loss": 0.9053,
      "step": 790
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00016870109546165884,
      "loss": 0.8741,
      "step": 800
    },
    {
      "epoch": 1.25,
      "eval_accuracy": 0.70508866615266,
      "eval_loss": 1.135292649269104,
      "eval_runtime": 42.8607,
      "eval_samples_per_second": 60.522,
      "eval_steps_per_second": 7.583,
      "step": 800
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00016830985915492958,
      "loss": 0.98,
      "step": 810
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00016791862284820032,
      "loss": 0.964,
      "step": 820
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00016752738654147106,
      "loss": 1.0007,
      "step": 830
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0001671361502347418,
      "loss": 0.7954,
      "step": 840
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00016674491392801253,
      "loss": 0.9806,
      "step": 850
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00016635367762128327,
      "loss": 0.8854,
      "step": 860
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00016596244131455401,
      "loss": 0.9212,
      "step": 870
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00016557120500782473,
      "loss": 0.8027,
      "step": 880
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00016517996870109547,
      "loss": 0.7901,
      "step": 890
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0001647887323943662,
      "loss": 0.8875,
      "step": 900
    },
    {
      "epoch": 1.41,
      "eval_accuracy": 0.7447956823438705,
      "eval_loss": 0.9951499700546265,
      "eval_runtime": 42.5213,
      "eval_samples_per_second": 61.005,
      "eval_steps_per_second": 7.643,
      "step": 900
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00016439749608763694,
      "loss": 0.8624,
      "step": 910
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.00016400625978090766,
      "loss": 0.7194,
      "step": 920
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0001636150234741784,
      "loss": 0.8012,
      "step": 930
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.00016322378716744916,
      "loss": 0.9272,
      "step": 940
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0001628325508607199,
      "loss": 0.8623,
      "step": 950
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00016244131455399061,
      "loss": 0.8576,
      "step": 960
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.00016205007824726135,
      "loss": 0.6199,
      "step": 970
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0001616588419405321,
      "loss": 0.6498,
      "step": 980
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.00016126760563380283,
      "loss": 0.8226,
      "step": 990
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.00016087636932707354,
      "loss": 0.7233,
      "step": 1000
    },
    {
      "epoch": 1.56,
      "eval_accuracy": 0.748650732459522,
      "eval_loss": 0.926507294178009,
      "eval_runtime": 43.2867,
      "eval_samples_per_second": 59.926,
      "eval_steps_per_second": 7.508,
      "step": 1000
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.00016048513302034428,
      "loss": 0.7018,
      "step": 1010
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.00016009389671361502,
      "loss": 0.739,
      "step": 1020
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.00015970266040688576,
      "loss": 0.6795,
      "step": 1030
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0001593114241001565,
      "loss": 0.6543,
      "step": 1040
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.00015892018779342724,
      "loss": 0.7367,
      "step": 1050
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.00015852895148669798,
      "loss": 0.6326,
      "step": 1060
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.00015813771517996872,
      "loss": 0.5459,
      "step": 1070
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.00015774647887323943,
      "loss": 0.8587,
      "step": 1080
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.00015735524256651017,
      "loss": 0.6657,
      "step": 1090
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.0001569640062597809,
      "loss": 0.6696,
      "step": 1100
    },
    {
      "epoch": 1.72,
      "eval_accuracy": 0.7625289128758674,
      "eval_loss": 0.8659528493881226,
      "eval_runtime": 42.5957,
      "eval_samples_per_second": 60.898,
      "eval_steps_per_second": 7.63,
      "step": 1100
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.00015657276995305165,
      "loss": 0.7146,
      "step": 1110
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.00015618153364632236,
      "loss": 0.4619,
      "step": 1120
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0001557902973395931,
      "loss": 0.6367,
      "step": 1130
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.00015539906103286387,
      "loss": 0.585,
      "step": 1140
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0001550078247261346,
      "loss": 0.5464,
      "step": 1150
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.00015461658841940532,
      "loss": 0.7695,
      "step": 1160
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.00015422535211267606,
      "loss": 0.7019,
      "step": 1170
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.0001538341158059468,
      "loss": 0.4508,
      "step": 1180
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.00015344287949921754,
      "loss": 0.6396,
      "step": 1190
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.00015305164319248828,
      "loss": 0.7364,
      "step": 1200
    },
    {
      "epoch": 1.88,
      "eval_accuracy": 0.7579028527370856,
      "eval_loss": 0.8710386753082275,
      "eval_runtime": 42.6201,
      "eval_samples_per_second": 60.863,
      "eval_steps_per_second": 7.626,
      "step": 1200
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.000152660406885759,
      "loss": 0.8301,
      "step": 1210
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.00015226917057902973,
      "loss": 0.6424,
      "step": 1220
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.00015187793427230047,
      "loss": 0.788,
      "step": 1230
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.0001514866979655712,
      "loss": 0.8115,
      "step": 1240
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.00015109546165884195,
      "loss": 0.7002,
      "step": 1250
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.0001507042253521127,
      "loss": 0.5162,
      "step": 1260
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.00015031298904538343,
      "loss": 0.5706,
      "step": 1270
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.00014992175273865417,
      "loss": 0.4565,
      "step": 1280
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.00014953051643192488,
      "loss": 0.3758,
      "step": 1290
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.00014913928012519562,
      "loss": 0.3933,
      "step": 1300
    },
    {
      "epoch": 2.03,
      "eval_accuracy": 0.8037779491133384,
      "eval_loss": 0.7161900997161865,
      "eval_runtime": 42.7187,
      "eval_samples_per_second": 60.723,
      "eval_steps_per_second": 7.608,
      "step": 1300
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.00014874804381846636,
      "loss": 0.2625,
      "step": 1310
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.0001483568075117371,
      "loss": 0.3585,
      "step": 1320
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.00014796557120500784,
      "loss": 0.3735,
      "step": 1330
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.00014757433489827858,
      "loss": 0.4221,
      "step": 1340
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.00014718309859154932,
      "loss": 0.3249,
      "step": 1350
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.00014679186228482006,
      "loss": 0.4802,
      "step": 1360
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.00014640062597809077,
      "loss": 0.3311,
      "step": 1370
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.0001460093896713615,
      "loss": 0.4276,
      "step": 1380
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.00014561815336463225,
      "loss": 0.222,
      "step": 1390
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.00014522691705790299,
      "loss": 0.3443,
      "step": 1400
    },
    {
      "epoch": 2.19,
      "eval_accuracy": 0.8299922898997687,
      "eval_loss": 0.6305217742919922,
      "eval_runtime": 43.2454,
      "eval_samples_per_second": 59.983,
      "eval_steps_per_second": 7.515,
      "step": 1400
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.0001448356807511737,
      "loss": 0.3609,
      "step": 1410
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.00014444444444444444,
      "loss": 0.4267,
      "step": 1420
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.00014405320813771518,
      "loss": 0.4206,
      "step": 1430
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.00014366197183098594,
      "loss": 0.3414,
      "step": 1440
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.00014327073552425666,
      "loss": 0.3351,
      "step": 1450
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.0001428794992175274,
      "loss": 0.2785,
      "step": 1460
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.00014248826291079813,
      "loss": 0.2454,
      "step": 1470
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.00014209702660406887,
      "loss": 0.4382,
      "step": 1480
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.0001417057902973396,
      "loss": 0.4117,
      "step": 1490
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.00014131455399061033,
      "loss": 0.3376,
      "step": 1500
    },
    {
      "epoch": 2.35,
      "eval_accuracy": 0.8315343099460293,
      "eval_loss": 0.62732994556427,
      "eval_runtime": 43.1453,
      "eval_samples_per_second": 60.122,
      "eval_steps_per_second": 7.533,
      "step": 1500
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.00014092331768388107,
      "loss": 0.4258,
      "step": 1510
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.0001405320813771518,
      "loss": 0.3927,
      "step": 1520
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.00014014084507042254,
      "loss": 0.4155,
      "step": 1530
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.00013974960876369328,
      "loss": 0.2391,
      "step": 1540
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.00013935837245696402,
      "loss": 0.2629,
      "step": 1550
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.00013896713615023476,
      "loss": 0.3786,
      "step": 1560
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.00013857589984350547,
      "loss": 0.3195,
      "step": 1570
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.00013818466353677621,
      "loss": 0.2218,
      "step": 1580
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.00013779342723004695,
      "loss": 0.2454,
      "step": 1590
    },
    {
      "epoch": 2.5,
      "learning_rate": 0.0001374021909233177,
      "loss": 0.3071,
      "step": 1600
    },
    {
      "epoch": 2.5,
      "eval_accuracy": 0.8319198149575945,
      "eval_loss": 0.5988024473190308,
      "eval_runtime": 43.1472,
      "eval_samples_per_second": 60.12,
      "eval_steps_per_second": 7.532,
      "step": 1600
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.0001370109546165884,
      "loss": 0.2436,
      "step": 1610
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.00013661971830985914,
      "loss": 0.244,
      "step": 1620
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.0001362284820031299,
      "loss": 0.3613,
      "step": 1630
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.00013583724569640065,
      "loss": 0.4596,
      "step": 1640
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.00013544600938967136,
      "loss": 0.2966,
      "step": 1650
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.0001350547730829421,
      "loss": 0.3579,
      "step": 1660
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.00013466353677621284,
      "loss": 0.3748,
      "step": 1670
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.00013427230046948358,
      "loss": 0.2644,
      "step": 1680
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.0001338810641627543,
      "loss": 0.2401,
      "step": 1690
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.00013348982785602503,
      "loss": 0.2863,
      "step": 1700
    },
    {
      "epoch": 2.66,
      "eval_accuracy": 0.8153430994602929,
      "eval_loss": 0.6730513572692871,
      "eval_runtime": 42.369,
      "eval_samples_per_second": 61.224,
      "eval_steps_per_second": 7.671,
      "step": 1700
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.00013309859154929577,
      "loss": 0.1875,
      "step": 1710
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.0001327073552425665,
      "loss": 0.2504,
      "step": 1720
    },
    {
      "epoch": 2.71,
      "learning_rate": 0.00013231611893583725,
      "loss": 0.3071,
      "step": 1730
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.000131924882629108,
      "loss": 0.3549,
      "step": 1740
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.00013153364632237873,
      "loss": 0.3535,
      "step": 1750
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.00013114241001564947,
      "loss": 0.3758,
      "step": 1760
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.0001307511737089202,
      "loss": 0.3174,
      "step": 1770
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.00013035993740219092,
      "loss": 0.2612,
      "step": 1780
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.00012996870109546166,
      "loss": 0.3562,
      "step": 1790
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.0001295774647887324,
      "loss": 0.3017,
      "step": 1800
    },
    {
      "epoch": 2.82,
      "eval_accuracy": 0.8315343099460293,
      "eval_loss": 0.6042011976242065,
      "eval_runtime": 43.2556,
      "eval_samples_per_second": 59.969,
      "eval_steps_per_second": 7.513,
      "step": 1800
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.00012918622848200314,
      "loss": 0.4068,
      "step": 1810
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.00012879499217527385,
      "loss": 0.2903,
      "step": 1820
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.00012840375586854462,
      "loss": 0.2385,
      "step": 1830
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.00012801251956181536,
      "loss": 0.2733,
      "step": 1840
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.0001276212832550861,
      "loss": 0.2174,
      "step": 1850
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.0001272300469483568,
      "loss": 0.2813,
      "step": 1860
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.00012683881064162755,
      "loss": 0.1874,
      "step": 1870
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.0001264475743348983,
      "loss": 0.3027,
      "step": 1880
    },
    {
      "epoch": 2.96,
      "learning_rate": 0.00012605633802816903,
      "loss": 0.3872,
      "step": 1890
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.00012566510172143974,
      "loss": 0.2382,
      "step": 1900
    },
    {
      "epoch": 2.97,
      "eval_accuracy": 0.8712413261372398,
      "eval_loss": 0.5117685794830322,
      "eval_runtime": 43.2378,
      "eval_samples_per_second": 59.994,
      "eval_steps_per_second": 7.517,
      "step": 1900
    },
    {
      "epoch": 2.99,
      "learning_rate": 0.00012527386541471048,
      "loss": 0.2799,
      "step": 1910
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.00012488262910798122,
      "loss": 0.2348,
      "step": 1920
    },
    {
      "epoch": 3.02,
      "learning_rate": 0.00012449139280125196,
      "loss": 0.1834,
      "step": 1930
    },
    {
      "epoch": 3.04,
      "learning_rate": 0.0001241001564945227,
      "loss": 0.2012,
      "step": 1940
    },
    {
      "epoch": 3.05,
      "learning_rate": 0.00012370892018779344,
      "loss": 0.182,
      "step": 1950
    },
    {
      "epoch": 3.07,
      "learning_rate": 0.00012331768388106418,
      "loss": 0.1467,
      "step": 1960
    },
    {
      "epoch": 3.08,
      "learning_rate": 0.00012292644757433492,
      "loss": 0.2158,
      "step": 1970
    },
    {
      "epoch": 3.1,
      "learning_rate": 0.00012253521126760563,
      "loss": 0.1666,
      "step": 1980
    },
    {
      "epoch": 3.11,
      "learning_rate": 0.00012214397496087637,
      "loss": 0.1223,
      "step": 1990
    },
    {
      "epoch": 3.13,
      "learning_rate": 0.00012175273865414711,
      "loss": 0.1578,
      "step": 2000
    },
    {
      "epoch": 3.13,
      "eval_accuracy": 0.8735543562066307,
      "eval_loss": 0.49165645241737366,
      "eval_runtime": 42.6596,
      "eval_samples_per_second": 60.807,
      "eval_steps_per_second": 7.618,
      "step": 2000
    },
    {
      "epoch": 3.15,
      "learning_rate": 0.00012136150234741786,
      "loss": 0.2114,
      "step": 2010
    },
    {
      "epoch": 3.16,
      "learning_rate": 0.00012097026604068857,
      "loss": 0.1657,
      "step": 2020
    },
    {
      "epoch": 3.18,
      "learning_rate": 0.00012057902973395931,
      "loss": 0.1228,
      "step": 2030
    },
    {
      "epoch": 3.19,
      "learning_rate": 0.00012018779342723005,
      "loss": 0.1246,
      "step": 2040
    },
    {
      "epoch": 3.21,
      "learning_rate": 0.00011979655712050079,
      "loss": 0.1231,
      "step": 2050
    },
    {
      "epoch": 3.22,
      "learning_rate": 0.00011940532081377152,
      "loss": 0.2011,
      "step": 2060
    },
    {
      "epoch": 3.24,
      "learning_rate": 0.00011901408450704226,
      "loss": 0.1177,
      "step": 2070
    },
    {
      "epoch": 3.26,
      "learning_rate": 0.000118622848200313,
      "loss": 0.1258,
      "step": 2080
    },
    {
      "epoch": 3.27,
      "learning_rate": 0.00011823161189358373,
      "loss": 0.1668,
      "step": 2090
    },
    {
      "epoch": 3.29,
      "learning_rate": 0.00011784037558685446,
      "loss": 0.1794,
      "step": 2100
    },
    {
      "epoch": 3.29,
      "eval_accuracy": 0.8631457208943716,
      "eval_loss": 0.53019118309021,
      "eval_runtime": 43.3542,
      "eval_samples_per_second": 59.833,
      "eval_steps_per_second": 7.496,
      "step": 2100
    },
    {
      "epoch": 3.3,
      "learning_rate": 0.0001174491392801252,
      "loss": 0.1714,
      "step": 2110
    },
    {
      "epoch": 3.32,
      "learning_rate": 0.00011705790297339594,
      "loss": 0.0935,
      "step": 2120
    },
    {
      "epoch": 3.33,
      "learning_rate": 0.00011666666666666668,
      "loss": 0.1422,
      "step": 2130
    },
    {
      "epoch": 3.35,
      "learning_rate": 0.0001162754303599374,
      "loss": 0.1965,
      "step": 2140
    },
    {
      "epoch": 3.36,
      "learning_rate": 0.00011588419405320814,
      "loss": 0.1014,
      "step": 2150
    },
    {
      "epoch": 3.38,
      "learning_rate": 0.00011549295774647888,
      "loss": 0.2306,
      "step": 2160
    },
    {
      "epoch": 3.4,
      "learning_rate": 0.00011510172143974962,
      "loss": 0.1482,
      "step": 2170
    },
    {
      "epoch": 3.41,
      "learning_rate": 0.00011471048513302034,
      "loss": 0.1835,
      "step": 2180
    },
    {
      "epoch": 3.43,
      "learning_rate": 0.00011431924882629107,
      "loss": 0.0886,
      "step": 2190
    },
    {
      "epoch": 3.44,
      "learning_rate": 0.00011392801251956183,
      "loss": 0.1093,
      "step": 2200
    },
    {
      "epoch": 3.44,
      "eval_accuracy": 0.8635312259059368,
      "eval_loss": 0.5034898519515991,
      "eval_runtime": 42.4689,
      "eval_samples_per_second": 61.08,
      "eval_steps_per_second": 7.653,
      "step": 2200
    },
    {
      "epoch": 3.46,
      "learning_rate": 0.00011353677621283257,
      "loss": 0.12,
      "step": 2210
    },
    {
      "epoch": 3.47,
      "learning_rate": 0.00011314553990610328,
      "loss": 0.1595,
      "step": 2220
    },
    {
      "epoch": 3.49,
      "learning_rate": 0.00011275430359937402,
      "loss": 0.1279,
      "step": 2230
    },
    {
      "epoch": 3.51,
      "learning_rate": 0.00011236306729264476,
      "loss": 0.1693,
      "step": 2240
    },
    {
      "epoch": 3.52,
      "learning_rate": 0.0001119718309859155,
      "loss": 0.1546,
      "step": 2250
    },
    {
      "epoch": 3.54,
      "learning_rate": 0.00011158059467918622,
      "loss": 0.0954,
      "step": 2260
    },
    {
      "epoch": 3.55,
      "learning_rate": 0.00011118935837245696,
      "loss": 0.1307,
      "step": 2270
    },
    {
      "epoch": 3.57,
      "learning_rate": 0.0001107981220657277,
      "loss": 0.1041,
      "step": 2280
    },
    {
      "epoch": 3.58,
      "learning_rate": 0.00011040688575899844,
      "loss": 0.0915,
      "step": 2290
    },
    {
      "epoch": 3.6,
      "learning_rate": 0.00011001564945226917,
      "loss": 0.1076,
      "step": 2300
    },
    {
      "epoch": 3.6,
      "eval_accuracy": 0.8673862760215882,
      "eval_loss": 0.5186420679092407,
      "eval_runtime": 42.5346,
      "eval_samples_per_second": 60.986,
      "eval_steps_per_second": 7.641,
      "step": 2300
    },
    {
      "epoch": 3.62,
      "learning_rate": 0.00010962441314553991,
      "loss": 0.1075,
      "step": 2310
    },
    {
      "epoch": 3.63,
      "learning_rate": 0.00010923317683881065,
      "loss": 0.0914,
      "step": 2320
    },
    {
      "epoch": 3.65,
      "learning_rate": 0.00010884194053208139,
      "loss": 0.1057,
      "step": 2330
    },
    {
      "epoch": 3.66,
      "learning_rate": 0.00010845070422535213,
      "loss": 0.1967,
      "step": 2340
    },
    {
      "epoch": 3.68,
      "learning_rate": 0.00010805946791862285,
      "loss": 0.1148,
      "step": 2350
    },
    {
      "epoch": 3.69,
      "learning_rate": 0.00010766823161189359,
      "loss": 0.0591,
      "step": 2360
    },
    {
      "epoch": 3.71,
      "learning_rate": 0.00010727699530516433,
      "loss": 0.1369,
      "step": 2370
    },
    {
      "epoch": 3.72,
      "learning_rate": 0.00010688575899843507,
      "loss": 0.0991,
      "step": 2380
    },
    {
      "epoch": 3.74,
      "learning_rate": 0.0001064945226917058,
      "loss": 0.1579,
      "step": 2390
    },
    {
      "epoch": 3.76,
      "learning_rate": 0.00010610328638497653,
      "loss": 0.1219,
      "step": 2400
    },
    {
      "epoch": 3.76,
      "eval_accuracy": 0.8801079414032382,
      "eval_loss": 0.4722863733768463,
      "eval_runtime": 42.9174,
      "eval_samples_per_second": 60.442,
      "eval_steps_per_second": 7.573,
      "step": 2400
    },
    {
      "epoch": 3.77,
      "learning_rate": 0.00010571205007824727,
      "loss": 0.0743,
      "step": 2410
    },
    {
      "epoch": 3.79,
      "learning_rate": 0.00010532081377151801,
      "loss": 0.1099,
      "step": 2420
    },
    {
      "epoch": 3.8,
      "learning_rate": 0.00010492957746478873,
      "loss": 0.1275,
      "step": 2430
    },
    {
      "epoch": 3.82,
      "learning_rate": 0.00010453834115805947,
      "loss": 0.0914,
      "step": 2440
    },
    {
      "epoch": 3.83,
      "learning_rate": 0.00010414710485133022,
      "loss": 0.1177,
      "step": 2450
    },
    {
      "epoch": 3.85,
      "learning_rate": 0.00010375586854460096,
      "loss": 0.1255,
      "step": 2460
    },
    {
      "epoch": 3.87,
      "learning_rate": 0.00010336463223787167,
      "loss": 0.1376,
      "step": 2470
    },
    {
      "epoch": 3.88,
      "learning_rate": 0.00010297339593114241,
      "loss": 0.1848,
      "step": 2480
    },
    {
      "epoch": 3.9,
      "learning_rate": 0.00010258215962441315,
      "loss": 0.0986,
      "step": 2490
    },
    {
      "epoch": 3.91,
      "learning_rate": 0.0001021909233176839,
      "loss": 0.1017,
      "step": 2500
    },
    {
      "epoch": 3.91,
      "eval_accuracy": 0.8712413261372398,
      "eval_loss": 0.5132078528404236,
      "eval_runtime": 42.9886,
      "eval_samples_per_second": 60.342,
      "eval_steps_per_second": 7.56,
      "step": 2500
    },
    {
      "epoch": 3.93,
      "learning_rate": 0.00010179968701095461,
      "loss": 0.1514,
      "step": 2510
    },
    {
      "epoch": 3.94,
      "learning_rate": 0.00010140845070422535,
      "loss": 0.0884,
      "step": 2520
    },
    {
      "epoch": 3.96,
      "learning_rate": 0.00010101721439749609,
      "loss": 0.0894,
      "step": 2530
    },
    {
      "epoch": 3.97,
      "learning_rate": 0.00010062597809076683,
      "loss": 0.0964,
      "step": 2540
    },
    {
      "epoch": 3.99,
      "learning_rate": 0.00010023474178403756,
      "loss": 0.0851,
      "step": 2550
    },
    {
      "epoch": 4.01,
      "learning_rate": 9.98435054773083e-05,
      "loss": 0.078,
      "step": 2560
    },
    {
      "epoch": 4.02,
      "learning_rate": 9.945226917057904e-05,
      "loss": 0.0611,
      "step": 2570
    },
    {
      "epoch": 4.04,
      "learning_rate": 9.906103286384976e-05,
      "loss": 0.0721,
      "step": 2580
    },
    {
      "epoch": 4.05,
      "learning_rate": 9.86697965571205e-05,
      "loss": 0.0604,
      "step": 2590
    },
    {
      "epoch": 4.07,
      "learning_rate": 9.827856025039124e-05,
      "loss": 0.0351,
      "step": 2600
    },
    {
      "epoch": 4.07,
      "eval_accuracy": 0.8727833461835004,
      "eval_loss": 0.4709137976169586,
      "eval_runtime": 43.1107,
      "eval_samples_per_second": 60.171,
      "eval_steps_per_second": 7.539,
      "step": 2600
    },
    {
      "epoch": 4.08,
      "learning_rate": 9.788732394366198e-05,
      "loss": 0.0262,
      "step": 2610
    },
    {
      "epoch": 4.1,
      "learning_rate": 9.749608763693271e-05,
      "loss": 0.0263,
      "step": 2620
    },
    {
      "epoch": 4.12,
      "learning_rate": 9.710485133020345e-05,
      "loss": 0.0201,
      "step": 2630
    },
    {
      "epoch": 4.13,
      "learning_rate": 9.671361502347419e-05,
      "loss": 0.0467,
      "step": 2640
    },
    {
      "epoch": 4.15,
      "learning_rate": 9.632237871674493e-05,
      "loss": 0.0635,
      "step": 2650
    },
    {
      "epoch": 4.16,
      "learning_rate": 9.593114241001565e-05,
      "loss": 0.0162,
      "step": 2660
    },
    {
      "epoch": 4.18,
      "learning_rate": 9.553990610328639e-05,
      "loss": 0.1138,
      "step": 2670
    },
    {
      "epoch": 4.19,
      "learning_rate": 9.514866979655712e-05,
      "loss": 0.0597,
      "step": 2680
    },
    {
      "epoch": 4.21,
      "learning_rate": 9.475743348982786e-05,
      "loss": 0.0349,
      "step": 2690
    },
    {
      "epoch": 4.23,
      "learning_rate": 9.43661971830986e-05,
      "loss": 0.0295,
      "step": 2700
    },
    {
      "epoch": 4.23,
      "eval_accuracy": 0.8824209714726291,
      "eval_loss": 0.4673934876918793,
      "eval_runtime": 43.0785,
      "eval_samples_per_second": 60.216,
      "eval_steps_per_second": 7.544,
      "step": 2700
    },
    {
      "epoch": 4.24,
      "learning_rate": 9.397496087636933e-05,
      "loss": 0.0375,
      "step": 2710
    },
    {
      "epoch": 4.26,
      "learning_rate": 9.358372456964006e-05,
      "loss": 0.0238,
      "step": 2720
    },
    {
      "epoch": 4.27,
      "learning_rate": 9.31924882629108e-05,
      "loss": 0.0262,
      "step": 2730
    },
    {
      "epoch": 4.29,
      "learning_rate": 9.280125195618154e-05,
      "loss": 0.0507,
      "step": 2740
    },
    {
      "epoch": 4.3,
      "learning_rate": 9.241001564945228e-05,
      "loss": 0.0397,
      "step": 2750
    },
    {
      "epoch": 4.32,
      "learning_rate": 9.2018779342723e-05,
      "loss": 0.0486,
      "step": 2760
    },
    {
      "epoch": 4.33,
      "learning_rate": 9.162754303599374e-05,
      "loss": 0.0332,
      "step": 2770
    },
    {
      "epoch": 4.35,
      "learning_rate": 9.123630672926447e-05,
      "loss": 0.022,
      "step": 2780
    },
    {
      "epoch": 4.37,
      "learning_rate": 9.084507042253522e-05,
      "loss": 0.0735,
      "step": 2790
    },
    {
      "epoch": 4.38,
      "learning_rate": 9.045383411580595e-05,
      "loss": 0.0416,
      "step": 2800
    },
    {
      "epoch": 4.38,
      "eval_accuracy": 0.8804934464148034,
      "eval_loss": 0.4836331307888031,
      "eval_runtime": 42.924,
      "eval_samples_per_second": 60.432,
      "eval_steps_per_second": 7.572,
      "step": 2800
    },
    {
      "epoch": 4.4,
      "learning_rate": 9.006259780907669e-05,
      "loss": 0.0188,
      "step": 2810
    },
    {
      "epoch": 4.41,
      "learning_rate": 8.967136150234741e-05,
      "loss": 0.0361,
      "step": 2820
    },
    {
      "epoch": 4.43,
      "learning_rate": 8.928012519561815e-05,
      "loss": 0.0655,
      "step": 2830
    },
    {
      "epoch": 4.44,
      "learning_rate": 8.888888888888889e-05,
      "loss": 0.097,
      "step": 2840
    },
    {
      "epoch": 4.46,
      "learning_rate": 8.849765258215963e-05,
      "loss": 0.1409,
      "step": 2850
    },
    {
      "epoch": 4.48,
      "learning_rate": 8.810641627543036e-05,
      "loss": 0.0353,
      "step": 2860
    },
    {
      "epoch": 4.49,
      "learning_rate": 8.77151799687011e-05,
      "loss": 0.036,
      "step": 2870
    },
    {
      "epoch": 4.51,
      "learning_rate": 8.732394366197182e-05,
      "loss": 0.0322,
      "step": 2880
    },
    {
      "epoch": 4.52,
      "learning_rate": 8.693270735524258e-05,
      "loss": 0.1002,
      "step": 2890
    },
    {
      "epoch": 4.54,
      "learning_rate": 8.65414710485133e-05,
      "loss": 0.0386,
      "step": 2900
    },
    {
      "epoch": 4.54,
      "eval_accuracy": 0.8828064764841943,
      "eval_loss": 0.466279000043869,
      "eval_runtime": 43.3083,
      "eval_samples_per_second": 59.896,
      "eval_steps_per_second": 7.504,
      "step": 2900
    },
    {
      "epoch": 4.55,
      "learning_rate": 8.615023474178404e-05,
      "loss": 0.0219,
      "step": 2910
    },
    {
      "epoch": 4.57,
      "learning_rate": 8.575899843505478e-05,
      "loss": 0.0428,
      "step": 2920
    },
    {
      "epoch": 4.59,
      "learning_rate": 8.536776212832551e-05,
      "loss": 0.0941,
      "step": 2930
    },
    {
      "epoch": 4.6,
      "learning_rate": 8.497652582159625e-05,
      "loss": 0.061,
      "step": 2940
    },
    {
      "epoch": 4.62,
      "learning_rate": 8.458528951486699e-05,
      "loss": 0.0524,
      "step": 2950
    },
    {
      "epoch": 4.63,
      "learning_rate": 8.419405320813773e-05,
      "loss": 0.0961,
      "step": 2960
    },
    {
      "epoch": 4.65,
      "learning_rate": 8.380281690140845e-05,
      "loss": 0.0524,
      "step": 2970
    },
    {
      "epoch": 4.66,
      "learning_rate": 8.341158059467919e-05,
      "loss": 0.0604,
      "step": 2980
    },
    {
      "epoch": 4.68,
      "learning_rate": 8.302034428794993e-05,
      "loss": 0.0534,
      "step": 2990
    },
    {
      "epoch": 4.69,
      "learning_rate": 8.262910798122067e-05,
      "loss": 0.0392,
      "step": 3000
    },
    {
      "epoch": 4.69,
      "eval_accuracy": 0.8989976869699307,
      "eval_loss": 0.4003235101699829,
      "eval_runtime": 43.2036,
      "eval_samples_per_second": 60.041,
      "eval_steps_per_second": 7.523,
      "step": 3000
    },
    {
      "epoch": 4.71,
      "learning_rate": 8.22378716744914e-05,
      "loss": 0.0273,
      "step": 3010
    },
    {
      "epoch": 4.73,
      "learning_rate": 8.184663536776213e-05,
      "loss": 0.0529,
      "step": 3020
    },
    {
      "epoch": 4.74,
      "learning_rate": 8.145539906103286e-05,
      "loss": 0.0726,
      "step": 3030
    },
    {
      "epoch": 4.76,
      "learning_rate": 8.106416275430361e-05,
      "loss": 0.0505,
      "step": 3040
    },
    {
      "epoch": 4.77,
      "learning_rate": 8.067292644757434e-05,
      "loss": 0.0778,
      "step": 3050
    },
    {
      "epoch": 4.79,
      "learning_rate": 8.028169014084508e-05,
      "loss": 0.0891,
      "step": 3060
    },
    {
      "epoch": 4.8,
      "learning_rate": 7.98904538341158e-05,
      "loss": 0.0264,
      "step": 3070
    },
    {
      "epoch": 4.82,
      "learning_rate": 7.949921752738654e-05,
      "loss": 0.0112,
      "step": 3080
    },
    {
      "epoch": 4.84,
      "learning_rate": 7.910798122065728e-05,
      "loss": 0.0163,
      "step": 3090
    },
    {
      "epoch": 4.85,
      "learning_rate": 7.871674491392802e-05,
      "loss": 0.0383,
      "step": 3100
    },
    {
      "epoch": 4.85,
      "eval_accuracy": 0.894757131842714,
      "eval_loss": 0.4187348186969757,
      "eval_runtime": 43.069,
      "eval_samples_per_second": 60.229,
      "eval_steps_per_second": 7.546,
      "step": 3100
    },
    {
      "epoch": 4.87,
      "learning_rate": 7.832550860719875e-05,
      "loss": 0.0867,
      "step": 3110
    },
    {
      "epoch": 4.88,
      "learning_rate": 7.793427230046949e-05,
      "loss": 0.0649,
      "step": 3120
    },
    {
      "epoch": 4.9,
      "learning_rate": 7.754303599374021e-05,
      "loss": 0.0959,
      "step": 3130
    },
    {
      "epoch": 4.91,
      "learning_rate": 7.715179968701097e-05,
      "loss": 0.0634,
      "step": 3140
    },
    {
      "epoch": 4.93,
      "learning_rate": 7.676056338028169e-05,
      "loss": 0.0607,
      "step": 3150
    },
    {
      "epoch": 4.95,
      "learning_rate": 7.636932707355243e-05,
      "loss": 0.0246,
      "step": 3160
    },
    {
      "epoch": 4.96,
      "learning_rate": 7.597809076682316e-05,
      "loss": 0.0421,
      "step": 3170
    },
    {
      "epoch": 4.98,
      "learning_rate": 7.55868544600939e-05,
      "loss": 0.0324,
      "step": 3180
    },
    {
      "epoch": 4.99,
      "learning_rate": 7.519561815336464e-05,
      "loss": 0.0119,
      "step": 3190
    },
    {
      "epoch": 5.01,
      "learning_rate": 7.480438184663538e-05,
      "loss": 0.0624,
      "step": 3200
    },
    {
      "epoch": 5.01,
      "eval_accuracy": 0.887432536622976,
      "eval_loss": 0.44595664739608765,
      "eval_runtime": 42.8626,
      "eval_samples_per_second": 60.519,
      "eval_steps_per_second": 7.582,
      "step": 3200
    },
    {
      "epoch": 5.02,
      "learning_rate": 7.44131455399061e-05,
      "loss": 0.0158,
      "step": 3210
    },
    {
      "epoch": 5.04,
      "learning_rate": 7.402190923317684e-05,
      "loss": 0.0264,
      "step": 3220
    },
    {
      "epoch": 5.05,
      "learning_rate": 7.363067292644758e-05,
      "loss": 0.0129,
      "step": 3230
    },
    {
      "epoch": 5.07,
      "learning_rate": 7.323943661971832e-05,
      "loss": 0.0153,
      "step": 3240
    },
    {
      "epoch": 5.09,
      "learning_rate": 7.284820031298905e-05,
      "loss": 0.0111,
      "step": 3250
    },
    {
      "epoch": 5.1,
      "learning_rate": 7.245696400625979e-05,
      "loss": 0.0179,
      "step": 3260
    },
    {
      "epoch": 5.12,
      "learning_rate": 7.206572769953051e-05,
      "loss": 0.0174,
      "step": 3270
    },
    {
      "epoch": 5.13,
      "learning_rate": 7.167449139280125e-05,
      "loss": 0.0396,
      "step": 3280
    },
    {
      "epoch": 5.15,
      "learning_rate": 7.128325508607199e-05,
      "loss": 0.0073,
      "step": 3290
    },
    {
      "epoch": 5.16,
      "learning_rate": 7.089201877934273e-05,
      "loss": 0.0188,
      "step": 3300
    },
    {
      "epoch": 5.16,
      "eval_accuracy": 0.9028527370855821,
      "eval_loss": 0.41694527864456177,
      "eval_runtime": 42.4867,
      "eval_samples_per_second": 61.054,
      "eval_steps_per_second": 7.649,
      "step": 3300
    },
    {
      "epoch": 5.18,
      "learning_rate": 7.050078247261346e-05,
      "loss": 0.0048,
      "step": 3310
    },
    {
      "epoch": 5.2,
      "learning_rate": 7.01095461658842e-05,
      "loss": 0.006,
      "step": 3320
    },
    {
      "epoch": 5.21,
      "learning_rate": 6.971830985915493e-05,
      "loss": 0.018,
      "step": 3330
    },
    {
      "epoch": 5.23,
      "learning_rate": 6.932707355242567e-05,
      "loss": 0.0314,
      "step": 3340
    },
    {
      "epoch": 5.24,
      "learning_rate": 6.89358372456964e-05,
      "loss": 0.0113,
      "step": 3350
    },
    {
      "epoch": 5.26,
      "learning_rate": 6.854460093896714e-05,
      "loss": 0.0075,
      "step": 3360
    },
    {
      "epoch": 5.27,
      "learning_rate": 6.815336463223787e-05,
      "loss": 0.0148,
      "step": 3370
    },
    {
      "epoch": 5.29,
      "learning_rate": 6.77621283255086e-05,
      "loss": 0.0092,
      "step": 3380
    },
    {
      "epoch": 5.31,
      "learning_rate": 6.737089201877934e-05,
      "loss": 0.022,
      "step": 3390
    },
    {
      "epoch": 5.32,
      "learning_rate": 6.697965571205008e-05,
      "loss": 0.0174,
      "step": 3400
    },
    {
      "epoch": 5.32,
      "eval_accuracy": 0.8951426368542791,
      "eval_loss": 0.40977469086647034,
      "eval_runtime": 42.8425,
      "eval_samples_per_second": 60.547,
      "eval_steps_per_second": 7.586,
      "step": 3400
    },
    {
      "epoch": 5.34,
      "learning_rate": 6.658841940532081e-05,
      "loss": 0.0054,
      "step": 3410
    },
    {
      "epoch": 5.35,
      "learning_rate": 6.619718309859155e-05,
      "loss": 0.0051,
      "step": 3420
    },
    {
      "epoch": 5.37,
      "learning_rate": 6.580594679186229e-05,
      "loss": 0.0316,
      "step": 3430
    },
    {
      "epoch": 5.38,
      "learning_rate": 6.541471048513303e-05,
      "loss": 0.0127,
      "step": 3440
    },
    {
      "epoch": 5.4,
      "learning_rate": 6.502347417840375e-05,
      "loss": 0.0085,
      "step": 3450
    },
    {
      "epoch": 5.41,
      "learning_rate": 6.463223787167449e-05,
      "loss": 0.0383,
      "step": 3460
    },
    {
      "epoch": 5.43,
      "learning_rate": 6.424100156494522e-05,
      "loss": 0.0075,
      "step": 3470
    },
    {
      "epoch": 5.45,
      "learning_rate": 6.384976525821597e-05,
      "loss": 0.0049,
      "step": 3480
    },
    {
      "epoch": 5.46,
      "learning_rate": 6.345852895148671e-05,
      "loss": 0.0259,
      "step": 3490
    },
    {
      "epoch": 5.48,
      "learning_rate": 6.306729264475744e-05,
      "loss": 0.0257,
      "step": 3500
    },
    {
      "epoch": 5.48,
      "eval_accuracy": 0.8951426368542791,
      "eval_loss": 0.4288833737373352,
      "eval_runtime": 42.534,
      "eval_samples_per_second": 60.987,
      "eval_steps_per_second": 7.641,
      "step": 3500
    },
    {
      "epoch": 5.49,
      "learning_rate": 6.267605633802818e-05,
      "loss": 0.0194,
      "step": 3510
    },
    {
      "epoch": 5.51,
      "learning_rate": 6.22848200312989e-05,
      "loss": 0.0144,
      "step": 3520
    },
    {
      "epoch": 5.52,
      "learning_rate": 6.189358372456964e-05,
      "loss": 0.0128,
      "step": 3530
    },
    {
      "epoch": 5.54,
      "learning_rate": 6.150234741784038e-05,
      "loss": 0.0273,
      "step": 3540
    },
    {
      "epoch": 5.56,
      "learning_rate": 6.111111111111112e-05,
      "loss": 0.0447,
      "step": 3550
    },
    {
      "epoch": 5.57,
      "learning_rate": 6.071987480438185e-05,
      "loss": 0.0389,
      "step": 3560
    },
    {
      "epoch": 5.59,
      "learning_rate": 6.032863849765259e-05,
      "loss": 0.0288,
      "step": 3570
    },
    {
      "epoch": 5.6,
      "learning_rate": 5.993740219092332e-05,
      "loss": 0.0101,
      "step": 3580
    },
    {
      "epoch": 5.62,
      "learning_rate": 5.954616588419406e-05,
      "loss": 0.006,
      "step": 3590
    },
    {
      "epoch": 5.63,
      "learning_rate": 5.915492957746479e-05,
      "loss": 0.0123,
      "step": 3600
    },
    {
      "epoch": 5.63,
      "eval_accuracy": 0.9028527370855821,
      "eval_loss": 0.4294571876525879,
      "eval_runtime": 42.3329,
      "eval_samples_per_second": 61.276,
      "eval_steps_per_second": 7.677,
      "step": 3600
    },
    {
      "epoch": 5.65,
      "learning_rate": 5.876369327073553e-05,
      "loss": 0.017,
      "step": 3610
    },
    {
      "epoch": 5.67,
      "learning_rate": 5.837245696400626e-05,
      "loss": 0.0132,
      "step": 3620
    },
    {
      "epoch": 5.68,
      "learning_rate": 5.7981220657277e-05,
      "loss": 0.0201,
      "step": 3630
    },
    {
      "epoch": 5.7,
      "learning_rate": 5.7589984350547735e-05,
      "loss": 0.0104,
      "step": 3640
    },
    {
      "epoch": 5.71,
      "learning_rate": 5.7198748043818474e-05,
      "loss": 0.0054,
      "step": 3650
    },
    {
      "epoch": 5.73,
      "learning_rate": 5.68075117370892e-05,
      "loss": 0.0137,
      "step": 3660
    },
    {
      "epoch": 5.74,
      "learning_rate": 5.6416275430359946e-05,
      "loss": 0.0038,
      "step": 3670
    },
    {
      "epoch": 5.76,
      "learning_rate": 5.602503912363067e-05,
      "loss": 0.0052,
      "step": 3680
    },
    {
      "epoch": 5.77,
      "learning_rate": 5.563380281690141e-05,
      "loss": 0.0142,
      "step": 3690
    },
    {
      "epoch": 5.79,
      "learning_rate": 5.5242566510172144e-05,
      "loss": 0.0052,
      "step": 3700
    },
    {
      "epoch": 5.79,
      "eval_accuracy": 0.8993831919814957,
      "eval_loss": 0.43951472640037537,
      "eval_runtime": 42.973,
      "eval_samples_per_second": 60.364,
      "eval_steps_per_second": 7.563,
      "step": 3700
    },
    {
      "epoch": 5.81,
      "learning_rate": 5.4851330203442884e-05,
      "loss": 0.0176,
      "step": 3710
    },
    {
      "epoch": 5.82,
      "learning_rate": 5.4460093896713616e-05,
      "loss": 0.0226,
      "step": 3720
    },
    {
      "epoch": 5.84,
      "learning_rate": 5.4068857589984356e-05,
      "loss": 0.007,
      "step": 3730
    },
    {
      "epoch": 5.85,
      "learning_rate": 5.367762128325509e-05,
      "loss": 0.0444,
      "step": 3740
    },
    {
      "epoch": 5.87,
      "learning_rate": 5.328638497652583e-05,
      "loss": 0.0056,
      "step": 3750
    },
    {
      "epoch": 5.88,
      "learning_rate": 5.2895148669796554e-05,
      "loss": 0.0172,
      "step": 3760
    },
    {
      "epoch": 5.9,
      "learning_rate": 5.25039123630673e-05,
      "loss": 0.0235,
      "step": 3770
    },
    {
      "epoch": 5.92,
      "learning_rate": 5.2112676056338026e-05,
      "loss": 0.0065,
      "step": 3780
    },
    {
      "epoch": 5.93,
      "learning_rate": 5.172143974960877e-05,
      "loss": 0.0083,
      "step": 3790
    },
    {
      "epoch": 5.95,
      "learning_rate": 5.13302034428795e-05,
      "loss": 0.0081,
      "step": 3800
    },
    {
      "epoch": 5.95,
      "eval_accuracy": 0.9082498072474943,
      "eval_loss": 0.42170995473861694,
      "eval_runtime": 43.0345,
      "eval_samples_per_second": 60.277,
      "eval_steps_per_second": 7.552,
      "step": 3800
    },
    {
      "epoch": 5.96,
      "learning_rate": 5.093896713615024e-05,
      "loss": 0.0071,
      "step": 3810
    },
    {
      "epoch": 5.98,
      "learning_rate": 5.054773082942097e-05,
      "loss": 0.031,
      "step": 3820
    },
    {
      "epoch": 5.99,
      "learning_rate": 5.015649452269171e-05,
      "loss": 0.0057,
      "step": 3830
    },
    {
      "epoch": 6.01,
      "learning_rate": 4.976525821596245e-05,
      "loss": 0.0051,
      "step": 3840
    },
    {
      "epoch": 6.03,
      "learning_rate": 4.937402190923318e-05,
      "loss": 0.0064,
      "step": 3850
    },
    {
      "epoch": 6.04,
      "learning_rate": 4.8982785602503914e-05,
      "loss": 0.0033,
      "step": 3860
    },
    {
      "epoch": 6.06,
      "learning_rate": 4.8591549295774653e-05,
      "loss": 0.0033,
      "step": 3870
    },
    {
      "epoch": 6.07,
      "learning_rate": 4.8200312989045386e-05,
      "loss": 0.0032,
      "step": 3880
    },
    {
      "epoch": 6.09,
      "learning_rate": 4.7809076682316126e-05,
      "loss": 0.0033,
      "step": 3890
    },
    {
      "epoch": 6.1,
      "learning_rate": 4.741784037558686e-05,
      "loss": 0.0032,
      "step": 3900
    },
    {
      "epoch": 6.1,
      "eval_accuracy": 0.9055512721665382,
      "eval_loss": 0.42163705825805664,
      "eval_runtime": 43.2354,
      "eval_samples_per_second": 59.997,
      "eval_steps_per_second": 7.517,
      "step": 3900
    },
    {
      "epoch": 6.12,
      "learning_rate": 4.702660406885759e-05,
      "loss": 0.0035,
      "step": 3910
    },
    {
      "epoch": 6.13,
      "learning_rate": 4.663536776212833e-05,
      "loss": 0.0032,
      "step": 3920
    },
    {
      "epoch": 6.15,
      "learning_rate": 4.624413145539906e-05,
      "loss": 0.003,
      "step": 3930
    },
    {
      "epoch": 6.17,
      "learning_rate": 4.58528951486698e-05,
      "loss": 0.0031,
      "step": 3940
    },
    {
      "epoch": 6.18,
      "learning_rate": 4.5461658841940535e-05,
      "loss": 0.0029,
      "step": 3950
    },
    {
      "epoch": 6.2,
      "learning_rate": 4.507042253521127e-05,
      "loss": 0.0115,
      "step": 3960
    },
    {
      "epoch": 6.21,
      "learning_rate": 4.467918622848201e-05,
      "loss": 0.0037,
      "step": 3970
    },
    {
      "epoch": 6.23,
      "learning_rate": 4.428794992175274e-05,
      "loss": 0.0056,
      "step": 3980
    },
    {
      "epoch": 6.24,
      "learning_rate": 4.389671361502348e-05,
      "loss": 0.0027,
      "step": 3990
    },
    {
      "epoch": 6.26,
      "learning_rate": 4.350547730829421e-05,
      "loss": 0.0033,
      "step": 4000
    },
    {
      "epoch": 6.26,
      "eval_accuracy": 0.9082498072474943,
      "eval_loss": 0.4112599790096283,
      "eval_runtime": 42.7966,
      "eval_samples_per_second": 60.612,
      "eval_steps_per_second": 7.594,
      "step": 4000
    },
    {
      "epoch": 6.28,
      "learning_rate": 4.311424100156495e-05,
      "loss": 0.0044,
      "step": 4010
    },
    {
      "epoch": 6.29,
      "learning_rate": 4.2723004694835684e-05,
      "loss": 0.0033,
      "step": 4020
    },
    {
      "epoch": 6.31,
      "learning_rate": 4.2331768388106416e-05,
      "loss": 0.0031,
      "step": 4030
    },
    {
      "epoch": 6.32,
      "learning_rate": 4.1940532081377156e-05,
      "loss": 0.0033,
      "step": 4040
    },
    {
      "epoch": 6.34,
      "learning_rate": 4.154929577464789e-05,
      "loss": 0.0048,
      "step": 4050
    },
    {
      "epoch": 6.35,
      "learning_rate": 4.115805946791863e-05,
      "loss": 0.003,
      "step": 4060
    },
    {
      "epoch": 6.37,
      "learning_rate": 4.076682316118936e-05,
      "loss": 0.0044,
      "step": 4070
    },
    {
      "epoch": 6.38,
      "learning_rate": 4.037558685446009e-05,
      "loss": 0.0037,
      "step": 4080
    },
    {
      "epoch": 6.4,
      "learning_rate": 3.998435054773083e-05,
      "loss": 0.0024,
      "step": 4090
    },
    {
      "epoch": 6.42,
      "learning_rate": 3.9593114241001565e-05,
      "loss": 0.0024,
      "step": 4100
    },
    {
      "epoch": 6.42,
      "eval_accuracy": 0.9101773323053199,
      "eval_loss": 0.40595710277557373,
      "eval_runtime": 42.4804,
      "eval_samples_per_second": 61.063,
      "eval_steps_per_second": 7.651,
      "step": 4100
    },
    {
      "epoch": 6.43,
      "learning_rate": 3.9201877934272305e-05,
      "loss": 0.0027,
      "step": 4110
    },
    {
      "epoch": 6.45,
      "learning_rate": 3.881064162754304e-05,
      "loss": 0.0036,
      "step": 4120
    },
    {
      "epoch": 6.46,
      "learning_rate": 3.841940532081377e-05,
      "loss": 0.0029,
      "step": 4130
    },
    {
      "epoch": 6.48,
      "learning_rate": 3.802816901408451e-05,
      "loss": 0.0057,
      "step": 4140
    },
    {
      "epoch": 6.49,
      "learning_rate": 3.763693270735524e-05,
      "loss": 0.0033,
      "step": 4150
    },
    {
      "epoch": 6.51,
      "learning_rate": 3.724569640062598e-05,
      "loss": 0.003,
      "step": 4160
    },
    {
      "epoch": 6.53,
      "learning_rate": 3.6854460093896714e-05,
      "loss": 0.0028,
      "step": 4170
    },
    {
      "epoch": 6.54,
      "learning_rate": 3.646322378716745e-05,
      "loss": 0.0028,
      "step": 4180
    },
    {
      "epoch": 6.56,
      "learning_rate": 3.6071987480438186e-05,
      "loss": 0.0075,
      "step": 4190
    },
    {
      "epoch": 6.57,
      "learning_rate": 3.568075117370892e-05,
      "loss": 0.0022,
      "step": 4200
    },
    {
      "epoch": 6.57,
      "eval_accuracy": 0.9090208172706246,
      "eval_loss": 0.4067469835281372,
      "eval_runtime": 42.6487,
      "eval_samples_per_second": 60.823,
      "eval_steps_per_second": 7.62,
      "step": 4200
    },
    {
      "epoch": 6.59,
      "learning_rate": 3.528951486697966e-05,
      "loss": 0.0046,
      "step": 4210
    },
    {
      "epoch": 6.6,
      "learning_rate": 3.489827856025039e-05,
      "loss": 0.0055,
      "step": 4220
    },
    {
      "epoch": 6.62,
      "learning_rate": 3.450704225352113e-05,
      "loss": 0.0026,
      "step": 4230
    },
    {
      "epoch": 6.64,
      "learning_rate": 3.411580594679186e-05,
      "loss": 0.0028,
      "step": 4240
    },
    {
      "epoch": 6.65,
      "learning_rate": 3.3724569640062596e-05,
      "loss": 0.005,
      "step": 4250
    },
    {
      "epoch": 6.67,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.0028,
      "step": 4260
    },
    {
      "epoch": 6.68,
      "learning_rate": 3.294209702660407e-05,
      "loss": 0.0023,
      "step": 4270
    },
    {
      "epoch": 6.7,
      "learning_rate": 3.255086071987481e-05,
      "loss": 0.0026,
      "step": 4280
    },
    {
      "epoch": 6.71,
      "learning_rate": 3.215962441314554e-05,
      "loss": 0.0025,
      "step": 4290
    },
    {
      "epoch": 6.73,
      "learning_rate": 3.176838810641627e-05,
      "loss": 0.0031,
      "step": 4300
    },
    {
      "epoch": 6.73,
      "eval_accuracy": 0.9113338473400154,
      "eval_loss": 0.4005250632762909,
      "eval_runtime": 43.1239,
      "eval_samples_per_second": 60.152,
      "eval_steps_per_second": 7.536,
      "step": 4300
    },
    {
      "epoch": 6.74,
      "learning_rate": 3.137715179968701e-05,
      "loss": 0.0027,
      "step": 4310
    },
    {
      "epoch": 6.76,
      "learning_rate": 3.0985915492957744e-05,
      "loss": 0.0024,
      "step": 4320
    },
    {
      "epoch": 6.78,
      "learning_rate": 3.0594679186228484e-05,
      "loss": 0.0031,
      "step": 4330
    },
    {
      "epoch": 6.79,
      "learning_rate": 3.0203442879499216e-05,
      "loss": 0.0109,
      "step": 4340
    },
    {
      "epoch": 6.81,
      "learning_rate": 2.9812206572769952e-05,
      "loss": 0.0032,
      "step": 4350
    },
    {
      "epoch": 6.82,
      "learning_rate": 2.9420970266040692e-05,
      "loss": 0.0027,
      "step": 4360
    },
    {
      "epoch": 6.84,
      "learning_rate": 2.9029733959311428e-05,
      "loss": 0.0033,
      "step": 4370
    },
    {
      "epoch": 6.85,
      "learning_rate": 2.8638497652582164e-05,
      "loss": 0.0044,
      "step": 4380
    },
    {
      "epoch": 6.87,
      "learning_rate": 2.82472613458529e-05,
      "loss": 0.0029,
      "step": 4390
    },
    {
      "epoch": 6.89,
      "learning_rate": 2.7856025039123636e-05,
      "loss": 0.0021,
      "step": 4400
    },
    {
      "epoch": 6.89,
      "eval_accuracy": 0.912875867386276,
      "eval_loss": 0.40083467960357666,
      "eval_runtime": 43.0173,
      "eval_samples_per_second": 60.301,
      "eval_steps_per_second": 7.555,
      "step": 4400
    },
    {
      "epoch": 6.9,
      "learning_rate": 2.746478873239437e-05,
      "loss": 0.0021,
      "step": 4410
    },
    {
      "epoch": 6.92,
      "learning_rate": 2.7073552425665105e-05,
      "loss": 0.0031,
      "step": 4420
    },
    {
      "epoch": 6.93,
      "learning_rate": 2.668231611893584e-05,
      "loss": 0.0023,
      "step": 4430
    },
    {
      "epoch": 6.95,
      "learning_rate": 2.6291079812206577e-05,
      "loss": 0.0032,
      "step": 4440
    },
    {
      "epoch": 6.96,
      "learning_rate": 2.5899843505477313e-05,
      "loss": 0.0339,
      "step": 4450
    },
    {
      "epoch": 6.98,
      "learning_rate": 2.5508607198748045e-05,
      "loss": 0.0025,
      "step": 4460
    },
    {
      "epoch": 7.0,
      "learning_rate": 2.511737089201878e-05,
      "loss": 0.0057,
      "step": 4470
    },
    {
      "epoch": 7.01,
      "learning_rate": 2.4726134585289514e-05,
      "loss": 0.0029,
      "step": 4480
    },
    {
      "epoch": 7.03,
      "learning_rate": 2.433489827856025e-05,
      "loss": 0.0019,
      "step": 4490
    },
    {
      "epoch": 7.04,
      "learning_rate": 2.3943661971830986e-05,
      "loss": 0.0021,
      "step": 4500
    },
    {
      "epoch": 7.04,
      "eval_accuracy": 0.9113338473400154,
      "eval_loss": 0.3966985046863556,
      "eval_runtime": 43.1598,
      "eval_samples_per_second": 60.102,
      "eval_steps_per_second": 7.53,
      "step": 4500
    },
    {
      "epoch": 7.06,
      "learning_rate": 2.3552425665101726e-05,
      "loss": 0.0034,
      "step": 4510
    },
    {
      "epoch": 7.07,
      "learning_rate": 2.3161189358372458e-05,
      "loss": 0.0023,
      "step": 4520
    },
    {
      "epoch": 7.09,
      "learning_rate": 2.2769953051643194e-05,
      "loss": 0.0025,
      "step": 4530
    },
    {
      "epoch": 7.1,
      "learning_rate": 2.237871674491393e-05,
      "loss": 0.0021,
      "step": 4540
    },
    {
      "epoch": 7.12,
      "learning_rate": 2.1987480438184666e-05,
      "loss": 0.0025,
      "step": 4550
    },
    {
      "epoch": 7.14,
      "learning_rate": 2.1596244131455402e-05,
      "loss": 0.0025,
      "step": 4560
    },
    {
      "epoch": 7.15,
      "learning_rate": 2.1205007824726135e-05,
      "loss": 0.0025,
      "step": 4570
    },
    {
      "epoch": 7.17,
      "learning_rate": 2.081377151799687e-05,
      "loss": 0.0023,
      "step": 4580
    },
    {
      "epoch": 7.18,
      "learning_rate": 2.0422535211267607e-05,
      "loss": 0.0023,
      "step": 4590
    },
    {
      "epoch": 7.2,
      "learning_rate": 2.0031298904538343e-05,
      "loss": 0.0043,
      "step": 4600
    },
    {
      "epoch": 7.2,
      "eval_accuracy": 0.9121048573631457,
      "eval_loss": 0.3959566652774811,
      "eval_runtime": 43.6577,
      "eval_samples_per_second": 59.417,
      "eval_steps_per_second": 7.444,
      "step": 4600
    },
    {
      "epoch": 7.21,
      "learning_rate": 1.964006259780908e-05,
      "loss": 0.0021,
      "step": 4610
    },
    {
      "epoch": 7.23,
      "learning_rate": 1.9248826291079812e-05,
      "loss": 0.0024,
      "step": 4620
    },
    {
      "epoch": 7.25,
      "learning_rate": 1.8857589984350548e-05,
      "loss": 0.0023,
      "step": 4630
    },
    {
      "epoch": 7.26,
      "learning_rate": 1.8466353677621284e-05,
      "loss": 0.0019,
      "step": 4640
    },
    {
      "epoch": 7.28,
      "learning_rate": 1.807511737089202e-05,
      "loss": 0.002,
      "step": 4650
    },
    {
      "epoch": 7.29,
      "learning_rate": 1.7683881064162756e-05,
      "loss": 0.0023,
      "step": 4660
    },
    {
      "epoch": 7.31,
      "learning_rate": 1.7292644757433492e-05,
      "loss": 0.0022,
      "step": 4670
    },
    {
      "epoch": 7.32,
      "learning_rate": 1.6901408450704224e-05,
      "loss": 0.0021,
      "step": 4680
    },
    {
      "epoch": 7.34,
      "learning_rate": 1.651017214397496e-05,
      "loss": 0.0021,
      "step": 4690
    },
    {
      "epoch": 7.36,
      "learning_rate": 1.6118935837245697e-05,
      "loss": 0.0022,
      "step": 4700
    },
    {
      "epoch": 7.36,
      "eval_accuracy": 0.9124903623747108,
      "eval_loss": 0.3962063193321228,
      "eval_runtime": 43.4452,
      "eval_samples_per_second": 59.707,
      "eval_steps_per_second": 7.481,
      "step": 4700
    },
    {
      "epoch": 7.37,
      "learning_rate": 1.5727699530516433e-05,
      "loss": 0.0023,
      "step": 4710
    },
    {
      "epoch": 7.39,
      "learning_rate": 1.533646322378717e-05,
      "loss": 0.0028,
      "step": 4720
    },
    {
      "epoch": 7.4,
      "learning_rate": 1.4945226917057903e-05,
      "loss": 0.0022,
      "step": 4730
    },
    {
      "epoch": 7.42,
      "learning_rate": 1.4553990610328639e-05,
      "loss": 0.0023,
      "step": 4740
    },
    {
      "epoch": 7.43,
      "learning_rate": 1.4162754303599373e-05,
      "loss": 0.0025,
      "step": 4750
    },
    {
      "epoch": 7.45,
      "learning_rate": 1.377151799687011e-05,
      "loss": 0.013,
      "step": 4760
    },
    {
      "epoch": 7.46,
      "learning_rate": 1.3380281690140845e-05,
      "loss": 0.0021,
      "step": 4770
    },
    {
      "epoch": 7.48,
      "learning_rate": 1.298904538341158e-05,
      "loss": 0.0023,
      "step": 4780
    },
    {
      "epoch": 7.5,
      "learning_rate": 1.2597809076682316e-05,
      "loss": 0.0027,
      "step": 4790
    },
    {
      "epoch": 7.51,
      "learning_rate": 1.2206572769953052e-05,
      "loss": 0.0021,
      "step": 4800
    },
    {
      "epoch": 7.51,
      "eval_accuracy": 0.9121048573631457,
      "eval_loss": 0.39916661381721497,
      "eval_runtime": 43.2453,
      "eval_samples_per_second": 59.983,
      "eval_steps_per_second": 7.515,
      "step": 4800
    },
    {
      "epoch": 7.53,
      "learning_rate": 1.1815336463223788e-05,
      "loss": 0.0021,
      "step": 4810
    },
    {
      "epoch": 7.54,
      "learning_rate": 1.1424100156494522e-05,
      "loss": 0.002,
      "step": 4820
    },
    {
      "epoch": 7.56,
      "learning_rate": 1.1032863849765258e-05,
      "loss": 0.0019,
      "step": 4830
    },
    {
      "epoch": 7.57,
      "learning_rate": 1.0641627543035994e-05,
      "loss": 0.0025,
      "step": 4840
    },
    {
      "epoch": 7.59,
      "learning_rate": 1.0250391236306729e-05,
      "loss": 0.0023,
      "step": 4850
    },
    {
      "epoch": 7.61,
      "learning_rate": 9.859154929577465e-06,
      "loss": 0.0022,
      "step": 4860
    },
    {
      "epoch": 7.62,
      "learning_rate": 9.4679186228482e-06,
      "loss": 0.0108,
      "step": 4870
    },
    {
      "epoch": 7.64,
      "learning_rate": 9.076682316118937e-06,
      "loss": 0.0021,
      "step": 4880
    },
    {
      "epoch": 7.65,
      "learning_rate": 8.685446009389673e-06,
      "loss": 0.0023,
      "step": 4890
    },
    {
      "epoch": 7.67,
      "learning_rate": 8.294209702660407e-06,
      "loss": 0.002,
      "step": 4900
    },
    {
      "epoch": 7.67,
      "eval_accuracy": 0.912875867386276,
      "eval_loss": 0.39511793851852417,
      "eval_runtime": 43.4397,
      "eval_samples_per_second": 59.715,
      "eval_steps_per_second": 7.482,
      "step": 4900
    },
    {
      "epoch": 7.68,
      "learning_rate": 7.902973395931143e-06,
      "loss": 0.0024,
      "step": 4910
    },
    {
      "epoch": 7.7,
      "learning_rate": 7.511737089201878e-06,
      "loss": 0.0021,
      "step": 4920
    },
    {
      "epoch": 7.72,
      "learning_rate": 7.120500782472614e-06,
      "loss": 0.0019,
      "step": 4930
    },
    {
      "epoch": 7.73,
      "learning_rate": 6.7292644757433494e-06,
      "loss": 0.0022,
      "step": 4940
    },
    {
      "epoch": 7.75,
      "learning_rate": 6.338028169014085e-06,
      "loss": 0.0023,
      "step": 4950
    },
    {
      "epoch": 7.76,
      "learning_rate": 5.94679186228482e-06,
      "loss": 0.002,
      "step": 4960
    },
    {
      "epoch": 7.78,
      "learning_rate": 5.555555555555556e-06,
      "loss": 0.0017,
      "step": 4970
    },
    {
      "epoch": 7.79,
      "learning_rate": 5.164319248826292e-06,
      "loss": 0.0024,
      "step": 4980
    },
    {
      "epoch": 7.81,
      "learning_rate": 4.773082942097027e-06,
      "loss": 0.0024,
      "step": 4990
    },
    {
      "epoch": 7.82,
      "learning_rate": 4.381846635367762e-06,
      "loss": 0.0023,
      "step": 5000
    },
    {
      "epoch": 7.82,
      "eval_accuracy": 0.9124903623747108,
      "eval_loss": 0.3951539695262909,
      "eval_runtime": 42.332,
      "eval_samples_per_second": 61.278,
      "eval_steps_per_second": 7.677,
      "step": 5000
    },
    {
      "epoch": 7.84,
      "learning_rate": 3.990610328638498e-06,
      "loss": 0.0021,
      "step": 5010
    },
    {
      "epoch": 7.86,
      "learning_rate": 3.5993740219092334e-06,
      "loss": 0.0025,
      "step": 5020
    },
    {
      "epoch": 7.87,
      "learning_rate": 3.2081377151799686e-06,
      "loss": 0.0029,
      "step": 5030
    },
    {
      "epoch": 7.89,
      "learning_rate": 2.8169014084507042e-06,
      "loss": 0.0024,
      "step": 5040
    },
    {
      "epoch": 7.9,
      "learning_rate": 2.42566510172144e-06,
      "loss": 0.0019,
      "step": 5050
    },
    {
      "epoch": 7.92,
      "learning_rate": 2.0344287949921754e-06,
      "loss": 0.0022,
      "step": 5060
    },
    {
      "epoch": 7.93,
      "learning_rate": 1.643192488262911e-06,
      "loss": 0.0025,
      "step": 5070
    },
    {
      "epoch": 7.95,
      "learning_rate": 1.2519561815336464e-06,
      "loss": 0.0018,
      "step": 5080
    },
    {
      "epoch": 7.97,
      "learning_rate": 8.607198748043818e-07,
      "loss": 0.0017,
      "step": 5090
    },
    {
      "epoch": 7.98,
      "learning_rate": 4.694835680751174e-07,
      "loss": 0.0021,
      "step": 5100
    },
    {
      "epoch": 7.98,
      "eval_accuracy": 0.912875867386276,
      "eval_loss": 0.39521878957748413,
      "eval_runtime": 43.0806,
      "eval_samples_per_second": 60.213,
      "eval_steps_per_second": 7.544,
      "step": 5100
    },
    {
      "epoch": 8.0,
      "learning_rate": 7.82472613458529e-08,
      "loss": 0.0026,
      "step": 5110
    },
    {
      "epoch": 8.0,
      "step": 5112,
      "total_flos": 6.337884979995771e+18,
      "train_loss": 0.4943873079753071,
      "train_runtime": 5596.588,
      "train_samples_per_second": 14.595,
      "train_steps_per_second": 0.913
    }
  ],
  "max_steps": 5112,
  "num_train_epochs": 8,
  "total_flos": 6.337884979995771e+18,
  "trial_name": null,
  "trial_params": null
}