{
  "best_metric": 0.9731557810188121,
  "best_model_checkpoint": "videomae-base-finetuned-ucf101-subset-V3KILLER\\checkpoint-4350",
  "epoch": 39.01845637583892,
  "eval_steps": 500,
  "global_step": 5960,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 8.389261744966443e-07,
      "loss": 0.4472,
      "step": 10
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.6778523489932886e-06,
      "loss": 0.4135,
      "step": 20
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.5167785234899326e-06,
      "loss": 0.7176,
      "step": 30
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.3557046979865773e-06,
      "loss": 0.5292,
      "step": 40
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.194630872483222e-06,
      "loss": 0.5718,
      "step": 50
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.033557046979865e-06,
      "loss": 0.6479,
      "step": 60
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.87248322147651e-06,
      "loss": 0.4323,
      "step": 70
    },
    {
      "epoch": 0.01,
      "learning_rate": 6.7114093959731546e-06,
      "loss": 0.6027,
      "step": 80
    },
    {
      "epoch": 0.02,
      "learning_rate": 7.5503355704698e-06,
      "loss": 0.3368,
      "step": 90
    },
    {
      "epoch": 0.02,
      "learning_rate": 8.389261744966444e-06,
      "loss": 0.4254,
      "step": 100
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.228187919463089e-06,
      "loss": 0.4692,
      "step": 110
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.006711409395973e-05,
      "loss": 0.3383,
      "step": 120
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.0906040268456376e-05,
      "loss": 0.5491,
      "step": 130
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.174496644295302e-05,
      "loss": 0.553,
      "step": 140
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.2583892617449666e-05,
      "loss": 0.3447,
      "step": 150
    },
    {
      "epoch": 0.03,
      "eval_accuracy": 0.9579370112027056,
      "eval_loss": 0.13385437428951263,
      "eval_runtime": 1375.006,
      "eval_samples_per_second": 3.441,
      "eval_steps_per_second": 0.431,
      "step": 150
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.3422818791946309e-05,
      "loss": 0.3939,
      "step": 160
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.4261744966442953e-05,
      "loss": 0.3293,
      "step": 170
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.51006711409396e-05,
      "loss": 0.4368,
      "step": 180
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.5939597315436243e-05,
      "loss": 0.3309,
      "step": 190
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.6778523489932888e-05,
      "loss": 0.4551,
      "step": 200
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.761744966442953e-05,
      "loss": 0.3551,
      "step": 210
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8456375838926178e-05,
      "loss": 0.5764,
      "step": 220
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.929530201342282e-05,
      "loss": 0.3553,
      "step": 230
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.013422818791946e-05,
      "loss": 0.4785,
      "step": 240
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.097315436241611e-05,
      "loss": 0.538,
      "step": 250
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.181208053691275e-05,
      "loss": 0.4879,
      "step": 260
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2651006711409396e-05,
      "loss": 0.3433,
      "step": 270
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.348993288590604e-05,
      "loss": 0.5569,
      "step": 280
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.4328859060402687e-05,
      "loss": 0.4604,
      "step": 290
    },
    {
      "epoch": 1.03,
      "learning_rate": 2.516778523489933e-05,
      "loss": 0.3161,
      "step": 300
    },
    {
      "epoch": 1.03,
      "eval_accuracy": 0.9465229338406257,
      "eval_loss": 0.15381179749965668,
      "eval_runtime": 1363.2798,
      "eval_samples_per_second": 3.47,
      "eval_steps_per_second": 0.434,
      "step": 300
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.6006711409395973e-05,
      "loss": 0.4349,
      "step": 310
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.6845637583892618e-05,
      "loss": 0.401,
      "step": 320
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.7684563758389263e-05,
      "loss": 0.5997,
      "step": 330
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.8523489932885905e-05,
      "loss": 0.5079,
      "step": 340
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.936241610738255e-05,
      "loss": 0.4201,
      "step": 350
    },
    {
      "epoch": 2.01,
      "learning_rate": 3.02013422818792e-05,
      "loss": 0.3062,
      "step": 360
    },
    {
      "epoch": 2.01,
      "learning_rate": 3.1040268456375844e-05,
      "loss": 0.3925,
      "step": 370
    },
    {
      "epoch": 2.01,
      "learning_rate": 3.1879194630872485e-05,
      "loss": 0.2427,
      "step": 380
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.271812080536913e-05,
      "loss": 0.4397,
      "step": 390
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.3557046979865775e-05,
      "loss": 0.4498,
      "step": 400
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.439597315436242e-05,
      "loss": 0.9242,
      "step": 410
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.523489932885906e-05,
      "loss": 0.4927,
      "step": 420
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.607382550335571e-05,
      "loss": 0.4862,
      "step": 430
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.6912751677852356e-05,
      "loss": 0.4986,
      "step": 440
    },
    {
      "epoch": 2.03,
      "learning_rate": 3.775167785234899e-05,
      "loss": 0.3386,
      "step": 450
    },
    {
      "epoch": 2.03,
      "eval_accuracy": 0.9019234834073134,
      "eval_loss": 0.3259753882884979,
      "eval_runtime": 1391.8832,
      "eval_samples_per_second": 3.399,
      "eval_steps_per_second": 0.425,
      "step": 450
    },
    {
      "epoch": 3.0,
      "learning_rate": 3.859060402684564e-05,
      "loss": 0.6401,
      "step": 460
    },
    {
      "epoch": 3.0,
      "learning_rate": 3.942953020134229e-05,
      "loss": 0.4998,
      "step": 470
    },
    {
      "epoch": 3.01,
      "learning_rate": 4.026845637583892e-05,
      "loss": 0.4991,
      "step": 480
    },
    {
      "epoch": 3.01,
      "learning_rate": 4.110738255033557e-05,
      "loss": 0.4428,
      "step": 490
    },
    {
      "epoch": 3.01,
      "learning_rate": 4.194630872483222e-05,
      "loss": 0.3152,
      "step": 500
    },
    {
      "epoch": 3.01,
      "learning_rate": 4.278523489932886e-05,
      "loss": 0.2307,
      "step": 510
    },
    {
      "epoch": 3.01,
      "learning_rate": 4.36241610738255e-05,
      "loss": 0.3793,
      "step": 520
    },
    {
      "epoch": 3.01,
      "learning_rate": 4.446308724832215e-05,
      "loss": 0.4842,
      "step": 530
    },
    {
      "epoch": 3.02,
      "learning_rate": 4.530201342281879e-05,
      "loss": 0.4256,
      "step": 540
    },
    {
      "epoch": 3.02,
      "learning_rate": 4.6140939597315434e-05,
      "loss": 0.2213,
      "step": 550
    },
    {
      "epoch": 3.02,
      "learning_rate": 4.697986577181208e-05,
      "loss": 0.8357,
      "step": 560
    },
    {
      "epoch": 3.02,
      "learning_rate": 4.7818791946308725e-05,
      "loss": 0.4367,
      "step": 570
    },
    {
      "epoch": 3.02,
      "learning_rate": 4.865771812080537e-05,
      "loss": 0.4018,
      "step": 580
    },
    {
      "epoch": 3.02,
      "learning_rate": 4.9496644295302015e-05,
      "loss": 0.4549,
      "step": 590
    },
    {
      "epoch": 3.03,
      "learning_rate": 4.99627143922446e-05,
      "loss": 0.3572,
      "step": 600
    },
    {
      "epoch": 3.03,
      "eval_accuracy": 0.9310927922215176,
      "eval_loss": 0.19672778248786926,
      "eval_runtime": 1380.4829,
      "eval_samples_per_second": 3.427,
      "eval_steps_per_second": 0.429,
      "step": 600
    },
    {
      "epoch": 4.0,
      "learning_rate": 4.9869500372856084e-05,
      "loss": 0.5104,
      "step": 610
    },
    {
      "epoch": 4.0,
      "learning_rate": 4.977628635346757e-05,
      "loss": 0.2492,
      "step": 620
    },
    {
      "epoch": 4.01,
      "learning_rate": 4.968307233407905e-05,
      "loss": 0.3431,
      "step": 630
    },
    {
      "epoch": 4.01,
      "learning_rate": 4.958985831469053e-05,
      "loss": 0.3546,
      "step": 640
    },
    {
      "epoch": 4.01,
      "learning_rate": 4.9496644295302015e-05,
      "loss": 0.5343,
      "step": 650
    },
    {
      "epoch": 4.01,
      "learning_rate": 4.94034302759135e-05,
      "loss": 0.4013,
      "step": 660
    },
    {
      "epoch": 4.01,
      "learning_rate": 4.931021625652498e-05,
      "loss": 0.4152,
      "step": 670
    },
    {
      "epoch": 4.01,
      "learning_rate": 4.921700223713647e-05,
      "loss": 0.3629,
      "step": 680
    },
    {
      "epoch": 4.02,
      "learning_rate": 4.912378821774795e-05,
      "loss": 0.4135,
      "step": 690
    },
    {
      "epoch": 4.02,
      "learning_rate": 4.9030574198359436e-05,
      "loss": 0.2992,
      "step": 700
    },
    {
      "epoch": 4.02,
      "learning_rate": 4.893736017897092e-05,
      "loss": 0.2103,
      "step": 710
    },
    {
      "epoch": 4.02,
      "learning_rate": 4.8844146159582404e-05,
      "loss": 0.4064,
      "step": 720
    },
    {
      "epoch": 4.02,
      "learning_rate": 4.875093214019389e-05,
      "loss": 0.3533,
      "step": 730
    },
    {
      "epoch": 4.02,
      "learning_rate": 4.865771812080537e-05,
      "loss": 0.5225,
      "step": 740
    },
    {
      "epoch": 4.03,
      "learning_rate": 4.856450410141686e-05,
      "loss": 0.3699,
      "step": 750
    },
    {
      "epoch": 4.03,
      "eval_accuracy": 0.9505389980976537,
      "eval_loss": 0.1660572737455368,
      "eval_runtime": 1390.2684,
      "eval_samples_per_second": 3.403,
      "eval_steps_per_second": 0.426,
      "step": 750
    },
    {
      "epoch": 5.0,
      "learning_rate": 4.8471290082028335e-05,
      "loss": 0.389,
      "step": 760
    },
    {
      "epoch": 5.0,
      "learning_rate": 4.837807606263982e-05,
      "loss": 0.3208,
      "step": 770
    },
    {
      "epoch": 5.01,
      "learning_rate": 4.82848620432513e-05,
      "loss": 0.2825,
      "step": 780
    },
    {
      "epoch": 5.01,
      "learning_rate": 4.819164802386279e-05,
      "loss": 0.3873,
      "step": 790
    },
    {
      "epoch": 5.01,
      "learning_rate": 4.809843400447427e-05,
      "loss": 0.1594,
      "step": 800
    },
    {
      "epoch": 5.01,
      "learning_rate": 4.8005219985085756e-05,
      "loss": 0.3303,
      "step": 810
    },
    {
      "epoch": 5.01,
      "learning_rate": 4.791200596569724e-05,
      "loss": 0.3552,
      "step": 820
    },
    {
      "epoch": 5.01,
      "learning_rate": 4.7818791946308725e-05,
      "loss": 0.3569,
      "step": 830
    },
    {
      "epoch": 5.02,
      "learning_rate": 4.772557792692021e-05,
      "loss": 0.5026,
      "step": 840
    },
    {
      "epoch": 5.02,
      "learning_rate": 4.763236390753169e-05,
      "loss": 0.2255,
      "step": 850
    },
    {
      "epoch": 5.02,
      "learning_rate": 4.753914988814318e-05,
      "loss": 0.2679,
      "step": 860
    },
    {
      "epoch": 5.02,
      "learning_rate": 4.744593586875467e-05,
      "loss": 0.3026,
      "step": 870
    },
    {
      "epoch": 5.02,
      "learning_rate": 4.735272184936615e-05,
      "loss": 0.3748,
      "step": 880
    },
    {
      "epoch": 5.02,
      "learning_rate": 4.725950782997763e-05,
      "loss": 0.811,
      "step": 890
    },
    {
      "epoch": 5.03,
      "learning_rate": 4.7166293810589114e-05,
      "loss": 0.3125,
      "step": 900
    },
    {
      "epoch": 5.03,
      "eval_accuracy": 0.9205242020714437,
      "eval_loss": 0.3291950523853302,
      "eval_runtime": 1436.751,
      "eval_samples_per_second": 3.293,
      "eval_steps_per_second": 0.412,
      "step": 900
    },
    {
      "epoch": 6.0,
      "learning_rate": 4.70730797912006e-05,
      "loss": 0.5823,
      "step": 910
    },
    {
      "epoch": 6.0,
      "learning_rate": 4.697986577181208e-05,
      "loss": 0.2634,
      "step": 920
    },
    {
      "epoch": 6.01,
      "learning_rate": 4.688665175242357e-05,
      "loss": 0.1815,
      "step": 930
    },
    {
      "epoch": 6.01,
      "learning_rate": 4.679343773303505e-05,
      "loss": 0.2191,
      "step": 940
    },
    {
      "epoch": 6.01,
      "learning_rate": 4.6700223713646536e-05,
      "loss": 0.301,
      "step": 950
    },
    {
      "epoch": 6.01,
      "learning_rate": 4.660700969425802e-05,
      "loss": 0.5404,
      "step": 960
    },
    {
      "epoch": 6.01,
      "learning_rate": 4.6513795674869504e-05,
      "loss": 0.4451,
      "step": 970
    },
    {
      "epoch": 6.01,
      "learning_rate": 4.642058165548099e-05,
      "loss": 0.1374,
      "step": 980
    },
    {
      "epoch": 6.02,
      "learning_rate": 4.632736763609247e-05,
      "loss": 0.3317,
      "step": 990
    },
    {
      "epoch": 6.02,
      "learning_rate": 4.623415361670396e-05,
      "loss": 0.2379,
      "step": 1000
    },
    {
      "epoch": 6.02,
      "learning_rate": 4.6140939597315434e-05,
      "loss": 0.4629,
      "step": 1010
    },
    {
      "epoch": 6.02,
      "learning_rate": 4.604772557792692e-05,
      "loss": 0.2929,
      "step": 1020
    },
    {
      "epoch": 6.02,
      "learning_rate": 4.59545115585384e-05,
      "loss": 0.3503,
      "step": 1030
    },
    {
      "epoch": 6.02,
      "learning_rate": 4.586129753914989e-05,
      "loss": 0.3551,
      "step": 1040
    },
    {
      "epoch": 6.03,
      "learning_rate": 4.576808351976137e-05,
      "loss": 0.4785,
      "step": 1050
    },
    {
      "epoch": 6.03,
      "eval_accuracy": 0.9323610230395265,
      "eval_loss": 0.2028670608997345,
      "eval_runtime": 1478.1245,
      "eval_samples_per_second": 3.201,
      "eval_steps_per_second": 0.401,
      "step": 1050
    },
    {
      "epoch": 7.0,
      "learning_rate": 4.5674869500372856e-05,
      "loss": 0.2218,
      "step": 1060
    },
    {
      "epoch": 7.0,
      "learning_rate": 4.558165548098434e-05,
      "loss": 0.2407,
      "step": 1070
    },
    {
      "epoch": 7.01,
      "learning_rate": 4.5488441461595824e-05,
      "loss": 0.4536,
      "step": 1080
    },
    {
      "epoch": 7.01,
      "learning_rate": 4.539522744220731e-05,
      "loss": 0.2698,
      "step": 1090
    },
    {
      "epoch": 7.01,
      "learning_rate": 4.530201342281879e-05,
      "loss": 0.2487,
      "step": 1100
    },
    {
      "epoch": 7.01,
      "learning_rate": 4.520879940343028e-05,
      "loss": 0.7459,
      "step": 1110
    },
    {
      "epoch": 7.01,
      "learning_rate": 4.511558538404176e-05,
      "loss": 0.3113,
      "step": 1120
    },
    {
      "epoch": 7.01,
      "learning_rate": 4.5022371364653246e-05,
      "loss": 0.1148,
      "step": 1130
    },
    {
      "epoch": 7.02,
      "learning_rate": 4.492915734526473e-05,
      "loss": 0.2549,
      "step": 1140
    },
    {
      "epoch": 7.02,
      "learning_rate": 4.4835943325876214e-05,
      "loss": 0.4447,
      "step": 1150
    },
    {
      "epoch": 7.02,
      "learning_rate": 4.47427293064877e-05,
      "loss": 0.5606,
      "step": 1160
    },
    {
      "epoch": 7.02,
      "learning_rate": 4.464951528709918e-05,
      "loss": 0.3264,
      "step": 1170
    },
    {
      "epoch": 7.02,
      "learning_rate": 4.455630126771067e-05,
      "loss": 0.2111,
      "step": 1180
    },
    {
      "epoch": 7.02,
      "learning_rate": 4.446308724832215e-05,
      "loss": 0.1777,
      "step": 1190
    },
    {
      "epoch": 7.03,
      "learning_rate": 4.4369873228933635e-05,
      "loss": 0.3477,
      "step": 1200
    },
    {
      "epoch": 7.03,
      "eval_accuracy": 0.9384908053265695,
      "eval_loss": 0.1534322202205658,
      "eval_runtime": 1491.9855,
      "eval_samples_per_second": 3.171,
      "eval_steps_per_second": 0.397,
      "step": 1200
    },
    {
      "epoch": 8.0,
      "learning_rate": 4.427665920954512e-05,
      "loss": 0.4373,
      "step": 1210
    },
    {
      "epoch": 8.0,
      "learning_rate": 4.4183445190156604e-05,
      "loss": 0.2638,
      "step": 1220
    },
    {
      "epoch": 8.01,
      "learning_rate": 4.409023117076809e-05,
      "loss": 0.3777,
      "step": 1230
    },
    {
      "epoch": 8.01,
      "learning_rate": 4.399701715137957e-05,
      "loss": 0.1847,
      "step": 1240
    },
    {
      "epoch": 8.01,
      "learning_rate": 4.390380313199106e-05,
      "loss": 0.3146,
      "step": 1250
    },
    {
      "epoch": 8.01,
      "learning_rate": 4.381058911260254e-05,
      "loss": 0.1038,
      "step": 1260
    },
    {
      "epoch": 8.01,
      "learning_rate": 4.371737509321402e-05,
      "loss": 0.27,
      "step": 1270
    },
    {
      "epoch": 8.01,
      "learning_rate": 4.36241610738255e-05,
      "loss": 0.3884,
      "step": 1280
    },
    {
      "epoch": 8.02,
      "learning_rate": 4.353094705443699e-05,
      "loss": 0.2249,
      "step": 1290
    },
    {
      "epoch": 8.02,
      "learning_rate": 4.343773303504847e-05,
      "loss": 0.2922,
      "step": 1300
    },
    {
      "epoch": 8.02,
      "learning_rate": 4.3344519015659955e-05,
      "loss": 0.2152,
      "step": 1310
    },
    {
      "epoch": 8.02,
      "learning_rate": 4.325130499627144e-05,
      "loss": 0.4288,
      "step": 1320
    },
    {
      "epoch": 8.02,
      "learning_rate": 4.3158090976882924e-05,
      "loss": 0.5197,
      "step": 1330
    },
    {
      "epoch": 8.02,
      "learning_rate": 4.306487695749441e-05,
      "loss": 0.1934,
      "step": 1340
    },
    {
      "epoch": 8.03,
      "learning_rate": 4.297166293810589e-05,
      "loss": 0.2909,
      "step": 1350
    },
    {
      "epoch": 8.03,
      "eval_accuracy": 0.9570915239906996,
      "eval_loss": 0.12647834420204163,
      "eval_runtime": 1464.2691,
      "eval_samples_per_second": 3.231,
      "eval_steps_per_second": 0.404,
      "step": 1350
    },
    {
      "epoch": 9.0,
      "learning_rate": 4.287844891871738e-05,
      "loss": 0.1825,
      "step": 1360
    },
    {
      "epoch": 9.0,
      "learning_rate": 4.278523489932886e-05,
      "loss": 0.2373,
      "step": 1370
    },
    {
      "epoch": 9.01,
      "learning_rate": 4.2692020879940345e-05,
      "loss": 0.3006,
      "step": 1380
    },
    {
      "epoch": 9.01,
      "learning_rate": 4.259880686055183e-05,
      "loss": 0.4427,
      "step": 1390
    },
    {
      "epoch": 9.01,
      "learning_rate": 4.2505592841163314e-05,
      "loss": 0.2385,
      "step": 1400
    },
    {
      "epoch": 9.01,
      "learning_rate": 4.24123788217748e-05,
      "loss": 0.2964,
      "step": 1410
    },
    {
      "epoch": 9.01,
      "learning_rate": 4.231916480238628e-05,
      "loss": 0.2207,
      "step": 1420
    },
    {
      "epoch": 9.01,
      "learning_rate": 4.2225950782997767e-05,
      "loss": 0.2755,
      "step": 1430
    },
    {
      "epoch": 9.02,
      "learning_rate": 4.213273676360925e-05,
      "loss": 0.1502,
      "step": 1440
    },
    {
      "epoch": 9.02,
      "learning_rate": 4.2039522744220735e-05,
      "loss": 0.1922,
      "step": 1450
    },
    {
      "epoch": 9.02,
      "learning_rate": 4.194630872483222e-05,
      "loss": 0.492,
      "step": 1460
    },
    {
      "epoch": 9.02,
      "learning_rate": 4.1853094705443704e-05,
      "loss": 0.5738,
      "step": 1470
    },
    {
      "epoch": 9.02,
      "learning_rate": 4.175988068605519e-05,
      "loss": 0.2227,
      "step": 1480
    },
    {
      "epoch": 9.02,
      "learning_rate": 4.166666666666667e-05,
      "loss": 0.3209,
      "step": 1490
    },
    {
      "epoch": 9.03,
      "learning_rate": 4.1573452647278156e-05,
      "loss": 0.2646,
      "step": 1500
    },
    {
      "epoch": 9.03,
      "eval_accuracy": 0.95857112661171,
      "eval_loss": 0.12387365847826004,
      "eval_runtime": 1486.6422,
      "eval_samples_per_second": 3.182,
      "eval_steps_per_second": 0.398,
      "step": 1500
    },
    {
      "epoch": 10.0,
      "learning_rate": 4.148023862788964e-05,
      "loss": 0.3465,
      "step": 1510
    },
    {
      "epoch": 10.0,
      "learning_rate": 4.138702460850112e-05,
      "loss": 0.3363,
      "step": 1520
    },
    {
      "epoch": 10.01,
      "learning_rate": 4.12938105891126e-05,
      "loss": 0.2512,
      "step": 1530
    },
    {
      "epoch": 10.01,
      "learning_rate": 4.1200596569724087e-05,
      "loss": 0.4622,
      "step": 1540
    },
    {
      "epoch": 10.01,
      "learning_rate": 4.110738255033557e-05,
      "loss": 0.2999,
      "step": 1550
    },
    {
      "epoch": 10.01,
      "learning_rate": 4.1014168530947055e-05,
      "loss": 0.2087,
      "step": 1560
    },
    {
      "epoch": 10.01,
      "learning_rate": 4.092095451155854e-05,
      "loss": 0.1872,
      "step": 1570
    },
    {
      "epoch": 10.01,
      "learning_rate": 4.0827740492170024e-05,
      "loss": 0.2439,
      "step": 1580
    },
    {
      "epoch": 10.02,
      "learning_rate": 4.073452647278151e-05,
      "loss": 0.279,
      "step": 1590
    },
    {
      "epoch": 10.02,
      "learning_rate": 4.064131245339299e-05,
      "loss": 0.3466,
      "step": 1600
    },
    {
      "epoch": 10.02,
      "learning_rate": 4.0548098434004476e-05,
      "loss": 0.167,
      "step": 1610
    },
    {
      "epoch": 10.02,
      "learning_rate": 4.045488441461596e-05,
      "loss": 0.3086,
      "step": 1620
    },
    {
      "epoch": 10.02,
      "learning_rate": 4.0361670395227445e-05,
      "loss": 0.1891,
      "step": 1630
    },
    {
      "epoch": 10.02,
      "learning_rate": 4.026845637583892e-05,
      "loss": 0.2303,
      "step": 1640
    },
    {
      "epoch": 10.03,
      "learning_rate": 4.017524235645041e-05,
      "loss": 0.3339,
      "step": 1650
    },
    {
      "epoch": 10.03,
      "eval_accuracy": 0.9627985626717396,
      "eval_loss": 0.13405466079711914,
      "eval_runtime": 1533.9642,
      "eval_samples_per_second": 3.084,
      "eval_steps_per_second": 0.386,
      "step": 1650
    },
    {
      "epoch": 11.0,
      "learning_rate": 4.00820283370619e-05,
      "loss": 0.1424,
      "step": 1660
    },
    {
      "epoch": 11.0,
      "learning_rate": 3.998881431767338e-05,
      "loss": 0.1715,
      "step": 1670
    },
    {
      "epoch": 11.01,
      "learning_rate": 3.9895600298284866e-05,
      "loss": 0.249,
      "step": 1680
    },
    {
      "epoch": 11.01,
      "learning_rate": 3.980238627889635e-05,
      "loss": 0.3207,
      "step": 1690
    },
    {
      "epoch": 11.01,
      "learning_rate": 3.9709172259507835e-05,
      "loss": 0.1715,
      "step": 1700
    },
    {
      "epoch": 11.01,
      "learning_rate": 3.961595824011932e-05,
      "loss": 0.163,
      "step": 1710
    },
    {
      "epoch": 11.01,
      "learning_rate": 3.95227442207308e-05,
      "loss": 0.1288,
      "step": 1720
    },
    {
      "epoch": 11.01,
      "learning_rate": 3.942953020134229e-05,
      "loss": 0.2297,
      "step": 1730
    },
    {
      "epoch": 11.02,
      "learning_rate": 3.933631618195377e-05,
      "loss": 0.6185,
      "step": 1740
    },
    {
      "epoch": 11.02,
      "learning_rate": 3.9243102162565256e-05,
      "loss": 0.2329,
      "step": 1750
    },
    {
      "epoch": 11.02,
      "learning_rate": 3.914988814317674e-05,
      "loss": 0.4634,
      "step": 1760
    },
    {
      "epoch": 11.02,
      "learning_rate": 3.905667412378822e-05,
      "loss": 0.2512,
      "step": 1770
    },
    {
      "epoch": 11.02,
      "learning_rate": 3.89634601043997e-05,
      "loss": 0.1862,
      "step": 1780
    },
    {
      "epoch": 11.02,
      "learning_rate": 3.8870246085011186e-05,
      "loss": 0.2681,
      "step": 1790
    },
    {
      "epoch": 11.03,
      "learning_rate": 3.877703206562267e-05,
      "loss": 0.0954,
      "step": 1800
    },
    {
      "epoch": 11.03,
      "eval_accuracy": 0.9422954977805961,
      "eval_loss": 0.18352869153022766,
      "eval_runtime": 1579.9144,
      "eval_samples_per_second": 2.994,
      "eval_steps_per_second": 0.375,
      "step": 1800
    },
    {
      "epoch": 12.0,
      "learning_rate": 3.8683818046234155e-05,
      "loss": 0.2858,
      "step": 1810
    },
    {
      "epoch": 12.0,
      "learning_rate": 3.859060402684564e-05,
      "loss": 0.1131,
      "step": 1820
    },
    {
      "epoch": 12.01,
      "learning_rate": 3.849739000745712e-05,
      "loss": 0.3666,
      "step": 1830
    },
    {
      "epoch": 12.01,
      "learning_rate": 3.840417598806861e-05,
      "loss": 0.0942,
      "step": 1840
    },
    {
      "epoch": 12.01,
      "learning_rate": 3.831096196868009e-05,
      "loss": 0.1977,
      "step": 1850
    },
    {
      "epoch": 12.01,
      "learning_rate": 3.8217747949291576e-05,
      "loss": 0.0735,
      "step": 1860
    },
    {
      "epoch": 12.01,
      "learning_rate": 3.812453392990306e-05,
      "loss": 0.3245,
      "step": 1870
    },
    {
      "epoch": 12.01,
      "learning_rate": 3.8031319910514545e-05,
      "loss": 0.125,
      "step": 1880
    },
    {
      "epoch": 12.02,
      "learning_rate": 3.793810589112602e-05,
      "loss": 0.1466,
      "step": 1890
    },
    {
      "epoch": 12.02,
      "learning_rate": 3.7844891871737506e-05,
      "loss": 0.1589,
      "step": 1900
    },
    {
      "epoch": 12.02,
      "learning_rate": 3.775167785234899e-05,
      "loss": 0.2833,
      "step": 1910
    },
    {
      "epoch": 12.02,
      "learning_rate": 3.7658463832960475e-05,
      "loss": 0.1586,
      "step": 1920
    },
    {
      "epoch": 12.02,
      "learning_rate": 3.756524981357196e-05,
      "loss": 0.1243,
      "step": 1930
    },
    {
      "epoch": 12.02,
      "learning_rate": 3.747203579418344e-05,
      "loss": 0.2151,
      "step": 1940
    },
    {
      "epoch": 12.03,
      "learning_rate": 3.737882177479493e-05,
      "loss": 0.3861,
      "step": 1950
    },
    {
      "epoch": 12.03,
      "eval_accuracy": 0.9467343056436271,
      "eval_loss": 0.22412604093551636,
      "eval_runtime": 1531.7023,
      "eval_samples_per_second": 3.089,
      "eval_steps_per_second": 0.386,
      "step": 1950
    },
    {
      "epoch": 13.0,
      "learning_rate": 3.728560775540642e-05,
      "loss": 0.2149,
      "step": 1960
    },
    {
      "epoch": 13.0,
      "learning_rate": 3.71923937360179e-05,
      "loss": 0.2922,
      "step": 1970
    },
    {
      "epoch": 13.01,
      "learning_rate": 3.709917971662939e-05,
      "loss": 0.0729,
      "step": 1980
    },
    {
      "epoch": 13.01,
      "learning_rate": 3.700596569724087e-05,
      "loss": 0.0935,
      "step": 1990
    },
    {
      "epoch": 13.01,
      "learning_rate": 3.6912751677852356e-05,
      "loss": 0.1765,
      "step": 2000
    },
    {
      "epoch": 13.01,
      "learning_rate": 3.681953765846384e-05,
      "loss": 0.3123,
      "step": 2010
    },
    {
      "epoch": 13.01,
      "learning_rate": 3.672632363907532e-05,
      "loss": 0.1471,
      "step": 2020
    },
    {
      "epoch": 13.01,
      "learning_rate": 3.66331096196868e-05,
      "loss": 0.3745,
      "step": 2030
    },
    {
      "epoch": 13.02,
      "learning_rate": 3.6539895600298286e-05,
      "loss": 0.0877,
      "step": 2040
    },
    {
      "epoch": 13.02,
      "learning_rate": 3.644668158090977e-05,
      "loss": 0.1444,
      "step": 2050
    },
    {
      "epoch": 13.02,
      "learning_rate": 3.6353467561521254e-05,
      "loss": 0.1732,
      "step": 2060
    },
    {
      "epoch": 13.02,
      "learning_rate": 3.626025354213274e-05,
      "loss": 0.3161,
      "step": 2070
    },
    {
      "epoch": 13.02,
      "learning_rate": 3.616703952274422e-05,
      "loss": 0.4199,
      "step": 2080
    },
    {
      "epoch": 13.02,
      "learning_rate": 3.607382550335571e-05,
      "loss": 0.3433,
      "step": 2090
    },
    {
      "epoch": 13.03,
      "learning_rate": 3.598061148396719e-05,
      "loss": 0.248,
      "step": 2100
    },
    {
      "epoch": 13.03,
      "eval_accuracy": 0.9619530754597336,
      "eval_loss": 0.12576867640018463,
      "eval_runtime": 1558.3572,
      "eval_samples_per_second": 3.036,
      "eval_steps_per_second": 0.38,
      "step": 2100
    },
    {
      "epoch": 14.0,
      "learning_rate": 3.5887397464578676e-05,
      "loss": 0.1326,
      "step": 2110
    },
    {
      "epoch": 14.0,
      "learning_rate": 3.579418344519016e-05,
      "loss": 0.1772,
      "step": 2120
    },
    {
      "epoch": 14.01,
      "learning_rate": 3.5700969425801644e-05,
      "loss": 0.3833,
      "step": 2130
    },
    {
      "epoch": 14.01,
      "learning_rate": 3.560775540641312e-05,
      "loss": 0.2131,
      "step": 2140
    },
    {
      "epoch": 14.01,
      "learning_rate": 3.5514541387024606e-05,
      "loss": 0.2458,
      "step": 2150
    },
    {
      "epoch": 14.01,
      "learning_rate": 3.542132736763609e-05,
      "loss": 0.0941,
      "step": 2160
    },
    {
      "epoch": 14.01,
      "learning_rate": 3.5328113348247575e-05,
      "loss": 0.128,
      "step": 2170
    },
    {
      "epoch": 14.01,
      "learning_rate": 3.523489932885906e-05,
      "loss": 0.0983,
      "step": 2180
    },
    {
      "epoch": 14.02,
      "learning_rate": 3.514168530947054e-05,
      "loss": 0.3903,
      "step": 2190
    },
    {
      "epoch": 14.02,
      "learning_rate": 3.504847129008203e-05,
      "loss": 0.1381,
      "step": 2200
    },
    {
      "epoch": 14.02,
      "learning_rate": 3.495525727069351e-05,
      "loss": 0.2005,
      "step": 2210
    },
    {
      "epoch": 14.02,
      "learning_rate": 3.4862043251304996e-05,
      "loss": 0.0698,
      "step": 2220
    },
    {
      "epoch": 14.02,
      "learning_rate": 3.476882923191648e-05,
      "loss": 0.1341,
      "step": 2230
    },
    {
      "epoch": 14.02,
      "learning_rate": 3.4675615212527964e-05,
      "loss": 0.0844,
      "step": 2240
    },
    {
      "epoch": 14.03,
      "learning_rate": 3.4582401193139455e-05,
      "loss": 0.2513,
      "step": 2250
    },
    {
      "epoch": 14.03,
      "eval_accuracy": 0.9357429718875502,
      "eval_loss": 0.2217186987400055,
      "eval_runtime": 1721.4439,
      "eval_samples_per_second": 2.748,
      "eval_steps_per_second": 0.344,
      "step": 2250
    },
    {
      "epoch": 15.0,
      "learning_rate": 3.448918717375094e-05,
      "loss": 0.1651,
      "step": 2260
    },
    {
      "epoch": 15.0,
      "learning_rate": 3.439597315436242e-05,
      "loss": 0.2301,
      "step": 2270
    },
    {
      "epoch": 15.01,
      "learning_rate": 3.43027591349739e-05,
      "loss": 0.0245,
      "step": 2280
    },
    {
      "epoch": 15.01,
      "learning_rate": 3.4209545115585386e-05,
      "loss": 0.0971,
      "step": 2290
    },
    {
      "epoch": 15.01,
      "learning_rate": 3.411633109619687e-05,
      "loss": 0.3628,
      "step": 2300
    },
    {
      "epoch": 15.01,
      "learning_rate": 3.4023117076808354e-05,
      "loss": 0.2239,
      "step": 2310
    },
    {
      "epoch": 15.01,
      "learning_rate": 3.392990305741984e-05,
      "loss": 0.022,
      "step": 2320
    },
    {
      "epoch": 15.01,
      "learning_rate": 3.383668903803132e-05,
      "loss": 0.2138,
      "step": 2330
    },
    {
      "epoch": 15.02,
      "learning_rate": 3.374347501864281e-05,
      "loss": 0.1416,
      "step": 2340
    },
    {
      "epoch": 15.02,
      "learning_rate": 3.365026099925429e-05,
      "loss": 0.1875,
      "step": 2350
    },
    {
      "epoch": 15.02,
      "learning_rate": 3.3557046979865775e-05,
      "loss": 0.1515,
      "step": 2360
    },
    {
      "epoch": 15.02,
      "learning_rate": 3.346383296047726e-05,
      "loss": 0.3692,
      "step": 2370
    },
    {
      "epoch": 15.02,
      "learning_rate": 3.3370618941088744e-05,
      "loss": 0.0893,
      "step": 2380
    },
    {
      "epoch": 15.02,
      "learning_rate": 3.327740492170022e-05,
      "loss": 0.0839,
      "step": 2390
    },
    {
      "epoch": 15.03,
      "learning_rate": 3.3184190902311706e-05,
      "loss": 0.1133,
      "step": 2400
    },
    {
      "epoch": 15.03,
      "eval_accuracy": 0.9406045233565842,
      "eval_loss": 0.21286079287528992,
      "eval_runtime": 1678.7607,
      "eval_samples_per_second": 2.818,
      "eval_steps_per_second": 0.353,
      "step": 2400
    },
    {
      "epoch": 16.0,
      "learning_rate": 3.309097688292319e-05,
      "loss": 0.1629,
      "step": 2410
    },
    {
      "epoch": 16.0,
      "learning_rate": 3.2997762863534674e-05,
      "loss": 0.0931,
      "step": 2420
    },
    {
      "epoch": 16.01,
      "learning_rate": 3.290454884414616e-05,
      "loss": 0.2746,
      "step": 2430
    },
    {
      "epoch": 16.01,
      "learning_rate": 3.281133482475764e-05,
      "loss": 0.0842,
      "step": 2440
    },
    {
      "epoch": 16.01,
      "learning_rate": 3.271812080536913e-05,
      "loss": 0.1122,
      "step": 2450
    },
    {
      "epoch": 16.01,
      "learning_rate": 3.262490678598061e-05,
      "loss": 0.0807,
      "step": 2460
    },
    {
      "epoch": 16.01,
      "learning_rate": 3.2531692766592095e-05,
      "loss": 0.3411,
      "step": 2470
    },
    {
      "epoch": 16.01,
      "learning_rate": 3.243847874720358e-05,
      "loss": 0.1466,
      "step": 2480
    },
    {
      "epoch": 16.02,
      "learning_rate": 3.2345264727815064e-05,
      "loss": 0.1806,
      "step": 2490
    },
    {
      "epoch": 16.02,
      "learning_rate": 3.225205070842655e-05,
      "loss": 0.1816,
      "step": 2500
    },
    {
      "epoch": 16.02,
      "learning_rate": 3.215883668903803e-05,
      "loss": 0.18,
      "step": 2510
    },
    {
      "epoch": 16.02,
      "learning_rate": 3.206562266964952e-05,
      "loss": 0.1056,
      "step": 2520
    },
    {
      "epoch": 16.02,
      "learning_rate": 3.1972408650261e-05,
      "loss": 0.0481,
      "step": 2530
    },
    {
      "epoch": 16.02,
      "learning_rate": 3.1879194630872485e-05,
      "loss": 0.2352,
      "step": 2540
    },
    {
      "epoch": 16.03,
      "learning_rate": 3.178598061148397e-05,
      "loss": 0.1421,
      "step": 2550
    },
    {
      "epoch": 16.03,
      "eval_accuracy": 0.9264426125554851,
      "eval_loss": 0.3005730211734772,
      "eval_runtime": 1661.3767,
      "eval_samples_per_second": 2.848,
      "eval_steps_per_second": 0.356,
      "step": 2550
    },
    {
      "epoch": 17.0,
      "learning_rate": 3.1692766592095454e-05,
      "loss": 0.0062,
      "step": 2560
    },
    {
      "epoch": 17.0,
      "learning_rate": 3.159955257270694e-05,
      "loss": 0.1642,
      "step": 2570
    },
    {
      "epoch": 17.01,
      "learning_rate": 3.150633855331842e-05,
      "loss": 0.1239,
      "step": 2580
    },
    {
      "epoch": 17.01,
      "learning_rate": 3.1413124533929907e-05,
      "loss": 0.1752,
      "step": 2590
    },
    {
      "epoch": 17.01,
      "learning_rate": 3.131991051454139e-05,
      "loss": 0.2545,
      "step": 2600
    },
    {
      "epoch": 17.01,
      "learning_rate": 3.1226696495152875e-05,
      "loss": 0.0558,
      "step": 2610
    },
    {
      "epoch": 17.01,
      "learning_rate": 3.113348247576436e-05,
      "loss": 0.1281,
      "step": 2620
    },
    {
      "epoch": 17.01,
      "learning_rate": 3.1040268456375844e-05,
      "loss": 0.0371,
      "step": 2630
    },
    {
      "epoch": 17.02,
      "learning_rate": 3.094705443698732e-05,
      "loss": 0.1842,
      "step": 2640
    },
    {
      "epoch": 17.02,
      "learning_rate": 3.0853840417598805e-05,
      "loss": 0.2683,
      "step": 2650
    },
    {
      "epoch": 17.02,
      "learning_rate": 3.076062639821029e-05,
      "loss": 0.1346,
      "step": 2660
    },
    {
      "epoch": 17.02,
      "learning_rate": 3.0667412378821774e-05,
      "loss": 0.0666,
      "step": 2670
    },
    {
      "epoch": 17.02,
      "learning_rate": 3.057419835943326e-05,
      "loss": 0.0707,
      "step": 2680
    },
    {
      "epoch": 17.02,
      "learning_rate": 3.0480984340044742e-05,
      "loss": 0.2043,
      "step": 2690
    },
    {
      "epoch": 17.03,
      "learning_rate": 3.0387770320656227e-05,
      "loss": 0.0248,
      "step": 2700
    },
    {
      "epoch": 17.03,
      "eval_accuracy": 0.9141830479813993,
      "eval_loss": 0.3867641091346741,
      "eval_runtime": 1615.8714,
      "eval_samples_per_second": 2.928,
      "eval_steps_per_second": 0.366,
      "step": 2700
    },
    {
      "epoch": 18.0,
      "learning_rate": 3.029455630126771e-05,
      "loss": 0.0546,
      "step": 2710
    },
    {
      "epoch": 18.0,
      "learning_rate": 3.02013422818792e-05,
      "loss": 0.2662,
      "step": 2720
    },
    {
      "epoch": 18.01,
      "learning_rate": 3.0108128262490683e-05,
      "loss": 0.1351,
      "step": 2730
    },
    {
      "epoch": 18.01,
      "learning_rate": 3.0014914243102167e-05,
      "loss": 0.1369,
      "step": 2740
    },
    {
      "epoch": 18.01,
      "learning_rate": 2.992170022371365e-05,
      "loss": 0.2072,
      "step": 2750
    },
    {
      "epoch": 18.01,
      "learning_rate": 2.9828486204325136e-05,
      "loss": 0.097,
      "step": 2760
    },
    {
      "epoch": 18.01,
      "learning_rate": 2.9735272184936613e-05,
      "loss": 0.1735,
      "step": 2770
    },
    {
      "epoch": 18.01,
      "learning_rate": 2.9642058165548097e-05,
      "loss": 0.2388,
      "step": 2780
    },
    {
      "epoch": 18.02,
      "learning_rate": 2.954884414615958e-05,
      "loss": 0.128,
      "step": 2790
    },
    {
      "epoch": 18.02,
      "learning_rate": 2.9455630126771066e-05,
      "loss": 0.0621,
      "step": 2800
    },
    {
      "epoch": 18.02,
      "learning_rate": 2.936241610738255e-05,
      "loss": 0.1235,
      "step": 2810
    },
    {
      "epoch": 18.02,
      "learning_rate": 2.9269202087994034e-05,
      "loss": 0.0467,
      "step": 2820
    },
    {
      "epoch": 18.02,
      "learning_rate": 2.917598806860552e-05,
      "loss": 0.1342,
      "step": 2830
    },
    {
      "epoch": 18.02,
      "learning_rate": 2.9082774049217003e-05,
      "loss": 0.0323,
      "step": 2840
    },
    {
      "epoch": 18.03,
      "learning_rate": 2.8989560029828487e-05,
      "loss": 0.0166,
      "step": 2850
    },
    {
      "epoch": 18.03,
      "eval_accuracy": 0.9518072289156626,
      "eval_loss": 0.25935444235801697,
      "eval_runtime": 1661.8912,
      "eval_samples_per_second": 2.847,
      "eval_steps_per_second": 0.356,
      "step": 2850
    },
    {
      "epoch": 19.0,
      "learning_rate": 2.889634601043997e-05,
      "loss": 0.0022,
      "step": 2860
    },
    {
      "epoch": 19.0,
      "learning_rate": 2.880313199105146e-05,
      "loss": 0.1903,
      "step": 2870
    },
    {
      "epoch": 19.01,
      "learning_rate": 2.8709917971662943e-05,
      "loss": 0.0789,
      "step": 2880
    },
    {
      "epoch": 19.01,
      "learning_rate": 2.861670395227442e-05,
      "loss": 0.3871,
      "step": 2890
    },
    {
      "epoch": 19.01,
      "learning_rate": 2.8523489932885905e-05,
      "loss": 0.0211,
      "step": 2900
    },
    {
      "epoch": 19.01,
      "learning_rate": 2.843027591349739e-05,
      "loss": 0.0834,
      "step": 2910
    },
    {
      "epoch": 19.01,
      "learning_rate": 2.8337061894108874e-05,
      "loss": 0.0381,
      "step": 2920
    },
    {
      "epoch": 19.01,
      "learning_rate": 2.8243847874720358e-05,
      "loss": 0.0987,
      "step": 2930
    },
    {
      "epoch": 19.02,
      "learning_rate": 2.8150633855331842e-05,
      "loss": 0.2549,
      "step": 2940
    },
    {
      "epoch": 19.02,
      "learning_rate": 2.8057419835943326e-05,
      "loss": 0.1535,
      "step": 2950
    },
    {
      "epoch": 19.02,
      "learning_rate": 2.796420581655481e-05,
      "loss": 0.001,
      "step": 2960
    },
    {
      "epoch": 19.02,
      "learning_rate": 2.7870991797166295e-05,
      "loss": 0.1714,
      "step": 2970
    },
    {
      "epoch": 19.02,
      "learning_rate": 2.777777777777778e-05,
      "loss": 0.1704,
      "step": 2980
    },
    {
      "epoch": 19.02,
      "learning_rate": 2.7684563758389263e-05,
      "loss": 0.0032,
      "step": 2990
    },
    {
      "epoch": 19.03,
      "learning_rate": 2.7591349739000748e-05,
      "loss": 0.0874,
      "step": 3000
    },
    {
      "epoch": 19.03,
      "eval_accuracy": 0.9251743817374762,
      "eval_loss": 0.365167498588562,
      "eval_runtime": 1786.4735,
      "eval_samples_per_second": 2.648,
      "eval_steps_per_second": 0.331,
      "step": 3000
    },
    {
      "epoch": 20.0,
      "learning_rate": 2.7498135719612232e-05,
      "loss": 0.0527,
      "step": 3010
    },
    {
      "epoch": 20.0,
      "learning_rate": 2.7404921700223713e-05,
      "loss": 0.122,
      "step": 3020
    },
    {
      "epoch": 20.01,
      "learning_rate": 2.7311707680835197e-05,
      "loss": 0.0586,
      "step": 3030
    },
    {
      "epoch": 20.01,
      "learning_rate": 2.721849366144668e-05,
      "loss": 0.0191,
      "step": 3040
    },
    {
      "epoch": 20.01,
      "learning_rate": 2.7125279642058166e-05,
      "loss": 0.0093,
      "step": 3050
    },
    {
      "epoch": 20.01,
      "learning_rate": 2.703206562266965e-05,
      "loss": 0.1019,
      "step": 3060
    },
    {
      "epoch": 20.01,
      "learning_rate": 2.6938851603281134e-05,
      "loss": 0.122,
      "step": 3070
    },
    {
      "epoch": 20.01,
      "learning_rate": 2.6845637583892618e-05,
      "loss": 0.3092,
      "step": 3080
    },
    {
      "epoch": 20.02,
      "learning_rate": 2.6752423564504103e-05,
      "loss": 0.1997,
      "step": 3090
    },
    {
      "epoch": 20.02,
      "learning_rate": 2.6659209545115587e-05,
      "loss": 0.2369,
      "step": 3100
    },
    {
      "epoch": 20.02,
      "learning_rate": 2.656599552572707e-05,
      "loss": 0.0795,
      "step": 3110
    },
    {
      "epoch": 20.02,
      "learning_rate": 2.6472781506338555e-05,
      "loss": 0.2416,
      "step": 3120
    },
    {
      "epoch": 20.02,
      "learning_rate": 2.637956748695004e-05,
      "loss": 0.0142,
      "step": 3130
    },
    {
      "epoch": 20.02,
      "learning_rate": 2.6286353467561524e-05,
      "loss": 0.0689,
      "step": 3140
    },
    {
      "epoch": 20.03,
      "learning_rate": 2.6193139448173005e-05,
      "loss": 0.0889,
      "step": 3150
    },
    {
      "epoch": 20.03,
      "eval_accuracy": 0.953286831536673,
      "eval_loss": 0.22492381930351257,
      "eval_runtime": 1711.3387,
      "eval_samples_per_second": 2.765,
      "eval_steps_per_second": 0.346,
      "step": 3150
    },
    {
      "epoch": 21.0,
      "learning_rate": 2.609992542878449e-05,
      "loss": 0.0716,
      "step": 3160
    },
    {
      "epoch": 21.0,
      "learning_rate": 2.6006711409395973e-05,
      "loss": 0.0802,
      "step": 3170
    },
    {
      "epoch": 21.01,
      "learning_rate": 2.5913497390007457e-05,
      "loss": 0.0101,
      "step": 3180
    },
    {
      "epoch": 21.01,
      "learning_rate": 2.5820283370618942e-05,
      "loss": 0.0745,
      "step": 3190
    },
    {
      "epoch": 21.01,
      "learning_rate": 2.5727069351230426e-05,
      "loss": 0.0442,
      "step": 3200
    },
    {
      "epoch": 21.01,
      "learning_rate": 2.563385533184191e-05,
      "loss": 0.0691,
      "step": 3210
    },
    {
      "epoch": 21.01,
      "learning_rate": 2.5540641312453395e-05,
      "loss": 0.0011,
      "step": 3220
    },
    {
      "epoch": 21.01,
      "learning_rate": 2.544742729306488e-05,
      "loss": 0.0725,
      "step": 3230
    },
    {
      "epoch": 21.02,
      "learning_rate": 2.5354213273676363e-05,
      "loss": 0.2005,
      "step": 3240
    },
    {
      "epoch": 21.02,
      "learning_rate": 2.5260999254287847e-05,
      "loss": 0.2762,
      "step": 3250
    },
    {
      "epoch": 21.02,
      "learning_rate": 2.516778523489933e-05,
      "loss": 0.0125,
      "step": 3260
    },
    {
      "epoch": 21.02,
      "learning_rate": 2.5074571215510812e-05,
      "loss": 0.0474,
      "step": 3270
    },
    {
      "epoch": 21.02,
      "learning_rate": 2.49813571961223e-05,
      "loss": 0.0579,
      "step": 3280
    },
    {
      "epoch": 21.02,
      "learning_rate": 2.4888143176733784e-05,
      "loss": 0.0021,
      "step": 3290
    },
    {
      "epoch": 21.03,
      "learning_rate": 2.4794929157345265e-05,
      "loss": 0.0804,
      "step": 3300
    },
    {
      "epoch": 21.03,
      "eval_accuracy": 0.9627985626717396,
      "eval_loss": 0.20273913443088531,
      "eval_runtime": 1680.7767,
      "eval_samples_per_second": 2.815,
      "eval_steps_per_second": 0.352,
      "step": 3300
    },
    {
      "epoch": 22.0,
      "learning_rate": 2.470171513795675e-05,
      "loss": 0.1415,
      "step": 3310
    },
    {
      "epoch": 22.0,
      "learning_rate": 2.4608501118568234e-05,
      "loss": 0.192,
      "step": 3320
    },
    {
      "epoch": 22.01,
      "learning_rate": 2.4515287099179718e-05,
      "loss": 0.0745,
      "step": 3330
    },
    {
      "epoch": 22.01,
      "learning_rate": 2.4422073079791202e-05,
      "loss": 0.3073,
      "step": 3340
    },
    {
      "epoch": 22.01,
      "learning_rate": 2.4328859060402687e-05,
      "loss": 0.1325,
      "step": 3350
    },
    {
      "epoch": 22.01,
      "learning_rate": 2.4235645041014167e-05,
      "loss": 0.0776,
      "step": 3360
    },
    {
      "epoch": 22.01,
      "learning_rate": 2.414243102162565e-05,
      "loss": 0.2178,
      "step": 3370
    },
    {
      "epoch": 22.01,
      "learning_rate": 2.4049217002237136e-05,
      "loss": 0.0093,
      "step": 3380
    },
    {
      "epoch": 22.02,
      "learning_rate": 2.395600298284862e-05,
      "loss": 0.2559,
      "step": 3390
    },
    {
      "epoch": 22.02,
      "learning_rate": 2.3862788963460104e-05,
      "loss": 0.1184,
      "step": 3400
    },
    {
      "epoch": 22.02,
      "learning_rate": 2.376957494407159e-05,
      "loss": 0.0537,
      "step": 3410
    },
    {
      "epoch": 22.02,
      "learning_rate": 2.3676360924683076e-05,
      "loss": 0.0965,
      "step": 3420
    },
    {
      "epoch": 22.02,
      "learning_rate": 2.3583146905294557e-05,
      "loss": 0.0402,
      "step": 3430
    },
    {
      "epoch": 22.02,
      "learning_rate": 2.348993288590604e-05,
      "loss": 0.0848,
      "step": 3440
    },
    {
      "epoch": 22.03,
      "learning_rate": 2.3396718866517526e-05,
      "loss": 0.0019,
      "step": 3450
    },
    {
      "epoch": 22.03,
      "eval_accuracy": 0.9211583174804481,
      "eval_loss": 0.4682135283946991,
      "eval_runtime": 1715.1321,
      "eval_samples_per_second": 2.758,
      "eval_steps_per_second": 0.345,
      "step": 3450
    },
    {
      "epoch": 23.0,
      "learning_rate": 2.330350484712901e-05,
      "loss": 0.0497,
      "step": 3460
    },
    {
      "epoch": 23.0,
      "learning_rate": 2.3210290827740494e-05,
      "loss": 0.0026,
      "step": 3470
    },
    {
      "epoch": 23.01,
      "learning_rate": 2.311707680835198e-05,
      "loss": 0.1634,
      "step": 3480
    },
    {
      "epoch": 23.01,
      "learning_rate": 2.302386278896346e-05,
      "loss": 0.1757,
      "step": 3490
    },
    {
      "epoch": 23.01,
      "learning_rate": 2.2930648769574944e-05,
      "loss": 0.0019,
      "step": 3500
    },
    {
      "epoch": 23.01,
      "learning_rate": 2.2837434750186428e-05,
      "loss": 0.0762,
      "step": 3510
    },
    {
      "epoch": 23.01,
      "learning_rate": 2.2744220730797912e-05,
      "loss": 0.0618,
      "step": 3520
    },
    {
      "epoch": 23.01,
      "learning_rate": 2.2651006711409396e-05,
      "loss": 0.2723,
      "step": 3530
    },
    {
      "epoch": 23.02,
      "learning_rate": 2.255779269202088e-05,
      "loss": 0.0711,
      "step": 3540
    },
    {
      "epoch": 23.02,
      "learning_rate": 2.2464578672632365e-05,
      "loss": 0.1,
      "step": 3550
    },
    {
      "epoch": 23.02,
      "learning_rate": 2.237136465324385e-05,
      "loss": 0.1396,
      "step": 3560
    },
    {
      "epoch": 23.02,
      "learning_rate": 2.2278150633855333e-05,
      "loss": 0.0402,
      "step": 3570
    },
    {
      "epoch": 23.02,
      "learning_rate": 2.2184936614466818e-05,
      "loss": 0.1067,
      "step": 3580
    },
    {
      "epoch": 23.02,
      "learning_rate": 2.2091722595078302e-05,
      "loss": 0.143,
      "step": 3590
    },
    {
      "epoch": 23.03,
      "learning_rate": 2.1998508575689786e-05,
      "loss": 0.0405,
      "step": 3600
    },
    {
      "epoch": 23.03,
      "eval_accuracy": 0.9492707672796449,
      "eval_loss": 0.2424902617931366,
      "eval_runtime": 1820.1723,
      "eval_samples_per_second": 2.599,
      "eval_steps_per_second": 0.325,
      "step": 3600
    },
    {
      "epoch": 24.0,
      "learning_rate": 2.190529455630127e-05,
      "loss": 0.0744,
      "step": 3610
    },
    {
      "epoch": 24.0,
      "learning_rate": 2.181208053691275e-05,
      "loss": 0.0584,
      "step": 3620
    },
    {
      "epoch": 24.01,
      "learning_rate": 2.1718866517524236e-05,
      "loss": 0.0455,
      "step": 3630
    },
    {
      "epoch": 24.01,
      "learning_rate": 2.162565249813572e-05,
      "loss": 0.1706,
      "step": 3640
    },
    {
      "epoch": 24.01,
      "learning_rate": 2.1532438478747204e-05,
      "loss": 0.1364,
      "step": 3650
    },
    {
      "epoch": 24.01,
      "learning_rate": 2.143922445935869e-05,
      "loss": 0.2008,
      "step": 3660
    },
    {
      "epoch": 24.01,
      "learning_rate": 2.1346010439970173e-05,
      "loss": 0.3287,
      "step": 3670
    },
    {
      "epoch": 24.01,
      "learning_rate": 2.1252796420581657e-05,
      "loss": 0.2396,
      "step": 3680
    },
    {
      "epoch": 24.02,
      "learning_rate": 2.115958240119314e-05,
      "loss": 0.1426,
      "step": 3690
    },
    {
      "epoch": 24.02,
      "learning_rate": 2.1066368381804625e-05,
      "loss": 0.002,
      "step": 3700
    },
    {
      "epoch": 24.02,
      "learning_rate": 2.097315436241611e-05,
      "loss": 0.0842,
      "step": 3710
    },
    {
      "epoch": 24.02,
      "learning_rate": 2.0879940343027594e-05,
      "loss": 0.1054,
      "step": 3720
    },
    {
      "epoch": 24.02,
      "learning_rate": 2.0786726323639078e-05,
      "loss": 0.236,
      "step": 3730
    },
    {
      "epoch": 24.02,
      "learning_rate": 2.069351230425056e-05,
      "loss": 0.0624,
      "step": 3740
    },
    {
      "epoch": 24.03,
      "learning_rate": 2.0600298284862043e-05,
      "loss": 0.0847,
      "step": 3750
    },
    {
      "epoch": 24.03,
      "eval_accuracy": 0.9558232931726908,
      "eval_loss": 0.2455640584230423,
      "eval_runtime": 1845.2522,
      "eval_samples_per_second": 2.564,
      "eval_steps_per_second": 0.321,
      "step": 3750
    },
    {
      "epoch": 25.0,
      "learning_rate": 2.0507084265473528e-05,
      "loss": 0.0553,
      "step": 3760
    },
    {
      "epoch": 25.0,
      "learning_rate": 2.0413870246085012e-05,
      "loss": 0.0602,
      "step": 3770
    },
    {
      "epoch": 25.01,
      "learning_rate": 2.0320656226696496e-05,
      "loss": 0.1579,
      "step": 3780
    },
    {
      "epoch": 25.01,
      "learning_rate": 2.022744220730798e-05,
      "loss": 0.0015,
      "step": 3790
    },
    {
      "epoch": 25.01,
      "learning_rate": 2.013422818791946e-05,
      "loss": 0.1018,
      "step": 3800
    },
    {
      "epoch": 25.01,
      "learning_rate": 2.004101416853095e-05,
      "loss": 0.0017,
      "step": 3810
    },
    {
      "epoch": 25.01,
      "learning_rate": 1.9947800149142433e-05,
      "loss": 0.0006,
      "step": 3820
    },
    {
      "epoch": 25.01,
      "learning_rate": 1.9854586129753917e-05,
      "loss": 0.0533,
      "step": 3830
    },
    {
      "epoch": 25.02,
      "learning_rate": 1.97613721103654e-05,
      "loss": 0.0886,
      "step": 3840
    },
    {
      "epoch": 25.02,
      "learning_rate": 1.9668158090976886e-05,
      "loss": 0.0006,
      "step": 3850
    },
    {
      "epoch": 25.02,
      "learning_rate": 1.957494407158837e-05,
      "loss": 0.1065,
      "step": 3860
    },
    {
      "epoch": 25.02,
      "learning_rate": 1.948173005219985e-05,
      "loss": 0.1181,
      "step": 3870
    },
    {
      "epoch": 25.02,
      "learning_rate": 1.9388516032811335e-05,
      "loss": 0.056,
      "step": 3880
    },
    {
      "epoch": 25.02,
      "learning_rate": 1.929530201342282e-05,
      "loss": 0.0875,
      "step": 3890
    },
    {
      "epoch": 25.03,
      "learning_rate": 1.9202087994034304e-05,
      "loss": 0.1656,
      "step": 3900
    },
    {
      "epoch": 25.03,
      "eval_accuracy": 0.9505389980976537,
      "eval_loss": 0.2623259723186493,
      "eval_runtime": 1799.7278,
      "eval_samples_per_second": 2.629,
      "eval_steps_per_second": 0.329,
      "step": 3900
    },
    {
      "epoch": 26.0,
      "learning_rate": 1.9108873974645788e-05,
      "loss": 0.0018,
      "step": 3910
    },
    {
      "epoch": 26.0,
      "learning_rate": 1.9015659955257272e-05,
      "loss": 0.1797,
      "step": 3920
    },
    {
      "epoch": 26.01,
      "learning_rate": 1.8922445935868753e-05,
      "loss": 0.1647,
      "step": 3930
    },
    {
      "epoch": 26.01,
      "learning_rate": 1.8829231916480237e-05,
      "loss": 0.0187,
      "step": 3940
    },
    {
      "epoch": 26.01,
      "learning_rate": 1.873601789709172e-05,
      "loss": 0.1639,
      "step": 3950
    },
    {
      "epoch": 26.01,
      "learning_rate": 1.864280387770321e-05,
      "loss": 0.0968,
      "step": 3960
    },
    {
      "epoch": 26.01,
      "learning_rate": 1.8549589858314694e-05,
      "loss": 0.0804,
      "step": 3970
    },
    {
      "epoch": 26.01,
      "learning_rate": 1.8456375838926178e-05,
      "loss": 0.0006,
      "step": 3980
    },
    {
      "epoch": 26.02,
      "learning_rate": 1.836316181953766e-05,
      "loss": 0.0542,
      "step": 3990
    },
    {
      "epoch": 26.02,
      "learning_rate": 1.8269947800149143e-05,
      "loss": 0.0006,
      "step": 4000
    },
    {
      "epoch": 26.02,
      "learning_rate": 1.8176733780760627e-05,
      "loss": 0.1271,
      "step": 4010
    },
    {
      "epoch": 26.02,
      "learning_rate": 1.808351976137211e-05,
      "loss": 0.1598,
      "step": 4020
    },
    {
      "epoch": 26.02,
      "learning_rate": 1.7990305741983596e-05,
      "loss": 0.0996,
      "step": 4030
    },
    {
      "epoch": 26.02,
      "learning_rate": 1.789709172259508e-05,
      "loss": 0.0736,
      "step": 4040
    },
    {
      "epoch": 26.03,
      "learning_rate": 1.780387770320656e-05,
      "loss": 0.1007,
      "step": 4050
    },
    {
      "epoch": 26.03,
      "eval_accuracy": 0.9484252800676389,
      "eval_loss": 0.23890338838100433,
      "eval_runtime": 1839.5641,
      "eval_samples_per_second": 2.572,
      "eval_steps_per_second": 0.322,
      "step": 4050
    },
    {
      "epoch": 27.0,
      "learning_rate": 1.7710663683818045e-05,
      "loss": 0.0617,
      "step": 4060
    },
    {
      "epoch": 27.0,
      "learning_rate": 1.761744966442953e-05,
      "loss": 0.0036,
      "step": 4070
    },
    {
      "epoch": 27.01,
      "learning_rate": 1.7524235645041014e-05,
      "loss": 0.0007,
      "step": 4080
    },
    {
      "epoch": 27.01,
      "learning_rate": 1.7431021625652498e-05,
      "loss": 0.0316,
      "step": 4090
    },
    {
      "epoch": 27.01,
      "learning_rate": 1.7337807606263982e-05,
      "loss": 0.0005,
      "step": 4100
    },
    {
      "epoch": 27.01,
      "learning_rate": 1.724459358687547e-05,
      "loss": 0.1614,
      "step": 4110
    },
    {
      "epoch": 27.01,
      "learning_rate": 1.715137956748695e-05,
      "loss": 0.0018,
      "step": 4120
    },
    {
      "epoch": 27.01,
      "learning_rate": 1.7058165548098435e-05,
      "loss": 0.1341,
      "step": 4130
    },
    {
      "epoch": 27.02,
      "learning_rate": 1.696495152870992e-05,
      "loss": 0.1544,
      "step": 4140
    },
    {
      "epoch": 27.02,
      "learning_rate": 1.6871737509321403e-05,
      "loss": 0.0413,
      "step": 4150
    },
    {
      "epoch": 27.02,
      "learning_rate": 1.6778523489932888e-05,
      "loss": 0.0015,
      "step": 4160
    },
    {
      "epoch": 27.02,
      "learning_rate": 1.6685309470544372e-05,
      "loss": 0.0592,
      "step": 4170
    },
    {
      "epoch": 27.02,
      "learning_rate": 1.6592095451155853e-05,
      "loss": 0.0037,
      "step": 4180
    },
    {
      "epoch": 27.02,
      "learning_rate": 1.6498881431767337e-05,
      "loss": 0.2047,
      "step": 4190
    },
    {
      "epoch": 27.03,
      "learning_rate": 1.640566741237882e-05,
      "loss": 0.0616,
      "step": 4200
    },
    {
      "epoch": 27.03,
      "eval_accuracy": 0.9543436905516804,
      "eval_loss": 0.25289884209632874,
      "eval_runtime": 1880.7167,
      "eval_samples_per_second": 2.516,
      "eval_steps_per_second": 0.315,
      "step": 4200
    },
    {
      "epoch": 28.0,
      "learning_rate": 1.6312453392990306e-05,
      "loss": 0.0006,
      "step": 4210
    },
    {
      "epoch": 28.0,
      "learning_rate": 1.621923937360179e-05,
      "loss": 0.1713,
      "step": 4220
    },
    {
      "epoch": 28.01,
      "learning_rate": 1.6126025354213274e-05,
      "loss": 0.1917,
      "step": 4230
    },
    {
      "epoch": 28.01,
      "learning_rate": 1.603281133482476e-05,
      "loss": 0.1431,
      "step": 4240
    },
    {
      "epoch": 28.01,
      "learning_rate": 1.5939597315436243e-05,
      "loss": 0.2054,
      "step": 4250
    },
    {
      "epoch": 28.01,
      "learning_rate": 1.5846383296047727e-05,
      "loss": 0.125,
      "step": 4260
    },
    {
      "epoch": 28.01,
      "learning_rate": 1.575316927665921e-05,
      "loss": 0.0045,
      "step": 4270
    },
    {
      "epoch": 28.01,
      "learning_rate": 1.5659955257270695e-05,
      "loss": 0.0736,
      "step": 4280
    },
    {
      "epoch": 28.02,
      "learning_rate": 1.556674123788218e-05,
      "loss": 0.0342,
      "step": 4290
    },
    {
      "epoch": 28.02,
      "learning_rate": 1.547352721849366e-05,
      "loss": 0.0841,
      "step": 4300
    },
    {
      "epoch": 28.02,
      "learning_rate": 1.5380313199105145e-05,
      "loss": 0.0819,
      "step": 4310
    },
    {
      "epoch": 28.02,
      "learning_rate": 1.528709917971663e-05,
      "loss": 0.001,
      "step": 4320
    },
    {
      "epoch": 28.02,
      "learning_rate": 1.5193885160328113e-05,
      "loss": 0.0652,
      "step": 4330
    },
    {
      "epoch": 28.02,
      "learning_rate": 1.51006711409396e-05,
      "loss": 0.0526,
      "step": 4340
    },
    {
      "epoch": 28.03,
      "learning_rate": 1.5007457121551084e-05,
      "loss": 0.0005,
      "step": 4350
    },
    {
      "epoch": 28.03,
      "eval_accuracy": 0.9731557810188121,
      "eval_loss": 0.152102530002594,
      "eval_runtime": 1850.8907,
      "eval_samples_per_second": 2.556,
      "eval_steps_per_second": 0.32,
      "step": 4350
    },
    {
      "epoch": 29.0,
      "learning_rate": 1.4914243102162568e-05,
      "loss": 0.0413,
      "step": 4360
    },
    {
      "epoch": 29.0,
      "learning_rate": 1.4821029082774049e-05,
      "loss": 0.0005,
      "step": 4370
    },
    {
      "epoch": 29.01,
      "learning_rate": 1.4727815063385533e-05,
      "loss": 0.003,
      "step": 4380
    },
    {
      "epoch": 29.01,
      "learning_rate": 1.4634601043997017e-05,
      "loss": 0.0672,
      "step": 4390
    },
    {
      "epoch": 29.01,
      "learning_rate": 1.4541387024608501e-05,
      "loss": 0.0878,
      "step": 4400
    },
    {
      "epoch": 29.01,
      "learning_rate": 1.4448173005219986e-05,
      "loss": 0.0715,
      "step": 4410
    },
    {
      "epoch": 29.01,
      "learning_rate": 1.4354958985831472e-05,
      "loss": 0.1653,
      "step": 4420
    },
    {
      "epoch": 29.01,
      "learning_rate": 1.4261744966442953e-05,
      "loss": 0.1316,
      "step": 4430
    },
    {
      "epoch": 29.02,
      "learning_rate": 1.4168530947054437e-05,
      "loss": 0.0678,
      "step": 4440
    },
    {
      "epoch": 29.02,
      "learning_rate": 1.4075316927665921e-05,
      "loss": 0.1543,
      "step": 4450
    },
    {
      "epoch": 29.02,
      "learning_rate": 1.3982102908277405e-05,
      "loss": 0.0025,
      "step": 4460
    },
    {
      "epoch": 29.02,
      "learning_rate": 1.388888888888889e-05,
      "loss": 0.0042,
      "step": 4470
    },
    {
      "epoch": 29.02,
      "learning_rate": 1.3795674869500374e-05,
      "loss": 0.0672,
      "step": 4480
    },
    {
      "epoch": 29.02,
      "learning_rate": 1.3702460850111856e-05,
      "loss": 0.0022,
      "step": 4490
    },
    {
      "epoch": 29.03,
      "learning_rate": 1.360924683072334e-05,
      "loss": 0.0006,
      "step": 4500
    },
    {
      "epoch": 29.03,
      "eval_accuracy": 0.9165081378144155,
      "eval_loss": 0.4115179777145386,
      "eval_runtime": 1860.5129,
      "eval_samples_per_second": 2.543,
      "eval_steps_per_second": 0.318,
      "step": 4500
    },
    {
      "epoch": 30.0,
      "learning_rate": 1.3516032811334825e-05,
      "loss": 0.0574,
      "step": 4510
    },
    {
      "epoch": 30.0,
      "learning_rate": 1.3422818791946309e-05,
      "loss": 0.0014,
      "step": 4520
    },
    {
      "epoch": 30.01,
      "learning_rate": 1.3329604772557793e-05,
      "loss": 0.0006,
      "step": 4530
    },
    {
      "epoch": 30.01,
      "learning_rate": 1.3236390753169278e-05,
      "loss": 0.0185,
      "step": 4540
    },
    {
      "epoch": 30.01,
      "learning_rate": 1.3143176733780762e-05,
      "loss": 0.0733,
      "step": 4550
    },
    {
      "epoch": 30.01,
      "learning_rate": 1.3049962714392244e-05,
      "loss": 0.1195,
      "step": 4560
    },
    {
      "epoch": 30.01,
      "learning_rate": 1.2956748695003729e-05,
      "loss": 0.0465,
      "step": 4570
    },
    {
      "epoch": 30.01,
      "learning_rate": 1.2863534675615213e-05,
      "loss": 0.0567,
      "step": 4580
    },
    {
      "epoch": 30.02,
      "learning_rate": 1.2770320656226697e-05,
      "loss": 0.0722,
      "step": 4590
    },
    {
      "epoch": 30.02,
      "learning_rate": 1.2677106636838182e-05,
      "loss": 0.0682,
      "step": 4600
    },
    {
      "epoch": 30.02,
      "learning_rate": 1.2583892617449666e-05,
      "loss": 0.0012,
      "step": 4610
    },
    {
      "epoch": 30.02,
      "learning_rate": 1.249067859806115e-05,
      "loss": 0.0455,
      "step": 4620
    },
    {
      "epoch": 30.02,
      "learning_rate": 1.2397464578672633e-05,
      "loss": 0.0004,
      "step": 4630
    },
    {
      "epoch": 30.02,
      "learning_rate": 1.2304250559284117e-05,
      "loss": 0.1246,
      "step": 4640
    },
    {
      "epoch": 30.03,
      "learning_rate": 1.2211036539895601e-05,
      "loss": 0.0007,
      "step": 4650
    },
    {
      "epoch": 30.03,
      "eval_accuracy": 0.922003804692454,
      "eval_loss": 0.4278564751148224,
      "eval_runtime": 1872.0845,
      "eval_samples_per_second": 2.527,
      "eval_steps_per_second": 0.316,
      "step": 4650
    },
    {
      "epoch": 31.0,
      "learning_rate": 1.2117822520507084e-05,
      "loss": 0.0979,
      "step": 4660
    },
    {
      "epoch": 31.0,
      "learning_rate": 1.2024608501118568e-05,
      "loss": 0.1198,
      "step": 4670
    },
    {
      "epoch": 31.01,
      "learning_rate": 1.1931394481730052e-05,
      "loss": 0.0525,
      "step": 4680
    },
    {
      "epoch": 31.01,
      "learning_rate": 1.1838180462341538e-05,
      "loss": 0.0005,
      "step": 4690
    },
    {
      "epoch": 31.01,
      "learning_rate": 1.174496644295302e-05,
      "loss": 0.0244,
      "step": 4700
    },
    {
      "epoch": 31.01,
      "learning_rate": 1.1651752423564505e-05,
      "loss": 0.0183,
      "step": 4710
    },
    {
      "epoch": 31.01,
      "learning_rate": 1.155853840417599e-05,
      "loss": 0.0033,
      "step": 4720
    },
    {
      "epoch": 31.01,
      "learning_rate": 1.1465324384787472e-05,
      "loss": 0.1106,
      "step": 4730
    },
    {
      "epoch": 31.02,
      "learning_rate": 1.1372110365398956e-05,
      "loss": 0.0528,
      "step": 4740
    },
    {
      "epoch": 31.02,
      "learning_rate": 1.127889634601044e-05,
      "loss": 0.0628,
      "step": 4750
    },
    {
      "epoch": 31.02,
      "learning_rate": 1.1185682326621925e-05,
      "loss": 0.0302,
      "step": 4760
    },
    {
      "epoch": 31.02,
      "learning_rate": 1.1092468307233409e-05,
      "loss": 0.0321,
      "step": 4770
    },
    {
      "epoch": 31.02,
      "learning_rate": 1.0999254287844893e-05,
      "loss": 0.0693,
      "step": 4780
    },
    {
      "epoch": 31.02,
      "learning_rate": 1.0906040268456376e-05,
      "loss": 0.1328,
      "step": 4790
    },
    {
      "epoch": 31.03,
      "learning_rate": 1.081282624906786e-05,
      "loss": 0.0004,
      "step": 4800
    },
    {
      "epoch": 31.03,
      "eval_accuracy": 0.9372225745085606,
      "eval_loss": 0.357156902551651,
      "eval_runtime": 1876.4105,
      "eval_samples_per_second": 2.521,
      "eval_steps_per_second": 0.315,
      "step": 4800
    },
    {
      "epoch": 32.0,
      "learning_rate": 1.0719612229679344e-05,
      "loss": 0.0483,
      "step": 4810
    },
    {
      "epoch": 32.0,
      "learning_rate": 1.0626398210290828e-05,
      "loss": 0.0291,
      "step": 4820
    },
    {
      "epoch": 32.01,
      "learning_rate": 1.0533184190902313e-05,
      "loss": 0.0757,
      "step": 4830
    },
    {
      "epoch": 32.01,
      "learning_rate": 1.0439970171513797e-05,
      "loss": 0.1715,
      "step": 4840
    },
    {
      "epoch": 32.01,
      "learning_rate": 1.034675615212528e-05,
      "loss": 0.0004,
      "step": 4850
    },
    {
      "epoch": 32.01,
      "learning_rate": 1.0253542132736764e-05,
      "loss": 0.0361,
      "step": 4860
    },
    {
      "epoch": 32.01,
      "learning_rate": 1.0160328113348248e-05,
      "loss": 0.0003,
      "step": 4870
    },
    {
      "epoch": 32.01,
      "learning_rate": 1.006711409395973e-05,
      "loss": 0.0117,
      "step": 4880
    },
    {
      "epoch": 32.02,
      "learning_rate": 9.973900074571217e-06,
      "loss": 0.0246,
      "step": 4890
    },
    {
      "epoch": 32.02,
      "learning_rate": 9.8806860551827e-06,
      "loss": 0.0015,
      "step": 4900
    },
    {
      "epoch": 32.02,
      "learning_rate": 9.787472035794185e-06,
      "loss": 0.0012,
      "step": 4910
    },
    {
      "epoch": 32.02,
      "learning_rate": 9.694258016405668e-06,
      "loss": 0.0002,
      "step": 4920
    },
    {
      "epoch": 32.02,
      "learning_rate": 9.601043997017152e-06,
      "loss": 0.0002,
      "step": 4930
    },
    {
      "epoch": 32.02,
      "learning_rate": 9.507829977628636e-06,
      "loss": 0.0003,
      "step": 4940
    },
    {
      "epoch": 32.03,
      "learning_rate": 9.414615958240119e-06,
      "loss": 0.0003,
      "step": 4950
    },
    {
      "epoch": 32.03,
      "eval_accuracy": 0.9418727541745932,
      "eval_loss": 0.3313732147216797,
      "eval_runtime": 1890.1725,
      "eval_samples_per_second": 2.503,
      "eval_steps_per_second": 0.313,
      "step": 4950
    },
    {
      "epoch": 33.0,
      "learning_rate": 9.321401938851605e-06,
      "loss": 0.0002,
      "step": 4960
    },
    {
      "epoch": 33.0,
      "learning_rate": 9.228187919463089e-06,
      "loss": 0.0487,
      "step": 4970
    },
    {
      "epoch": 33.01,
      "learning_rate": 9.134973900074571e-06,
      "loss": 0.0004,
      "step": 4980
    },
    {
      "epoch": 33.01,
      "learning_rate": 9.041759880686056e-06,
      "loss": 0.0779,
      "step": 4990
    },
    {
      "epoch": 33.01,
      "learning_rate": 8.94854586129754e-06,
      "loss": 0.053,
      "step": 5000
    },
    {
      "epoch": 33.01,
      "learning_rate": 8.855331841909023e-06,
      "loss": 0.0003,
      "step": 5010
    },
    {
      "epoch": 33.01,
      "learning_rate": 8.762117822520507e-06,
      "loss": 0.0002,
      "step": 5020
    },
    {
      "epoch": 33.01,
      "learning_rate": 8.668903803131991e-06,
      "loss": 0.0002,
      "step": 5030
    },
    {
      "epoch": 33.02,
      "learning_rate": 8.575689783743475e-06,
      "loss": 0.0032,
      "step": 5040
    },
    {
      "epoch": 33.02,
      "learning_rate": 8.48247576435496e-06,
      "loss": 0.0001,
      "step": 5050
    },
    {
      "epoch": 33.02,
      "learning_rate": 8.389261744966444e-06,
      "loss": 0.07,
      "step": 5060
    },
    {
      "epoch": 33.02,
      "learning_rate": 8.296047725577926e-06,
      "loss": 0.0032,
      "step": 5070
    },
    {
      "epoch": 33.02,
      "learning_rate": 8.20283370618941e-06,
      "loss": 0.0005,
      "step": 5080
    },
    {
      "epoch": 33.02,
      "learning_rate": 8.109619686800895e-06,
      "loss": 0.0138,
      "step": 5090
    },
    {
      "epoch": 33.03,
      "learning_rate": 8.01640566741238e-06,
      "loss": 0.0002,
      "step": 5100
    },
    {
      "epoch": 33.03,
      "eval_accuracy": 0.9346861128725428,
      "eval_loss": 0.40076977014541626,
      "eval_runtime": 1773.6137,
      "eval_samples_per_second": 2.667,
      "eval_steps_per_second": 0.334,
      "step": 5100
    },
    {
      "epoch": 34.0,
      "learning_rate": 7.923191648023863e-06,
      "loss": 0.054,
      "step": 5110
    },
    {
      "epoch": 34.0,
      "learning_rate": 7.829977628635348e-06,
      "loss": 0.0119,
      "step": 5120
    },
    {
      "epoch": 34.01,
      "learning_rate": 7.73676360924683e-06,
      "loss": 0.0178,
      "step": 5130
    },
    {
      "epoch": 34.01,
      "learning_rate": 7.643549589858315e-06,
      "loss": 0.0002,
      "step": 5140
    },
    {
      "epoch": 34.01,
      "learning_rate": 7.5503355704698e-06,
      "loss": 0.0956,
      "step": 5150
    },
    {
      "epoch": 34.01,
      "learning_rate": 7.457121551081284e-06,
      "loss": 0.0004,
      "step": 5160
    },
    {
      "epoch": 34.01,
      "learning_rate": 7.3639075316927665e-06,
      "loss": 0.0006,
      "step": 5170
    },
    {
      "epoch": 34.01,
      "learning_rate": 7.270693512304251e-06,
      "loss": 0.2637,
      "step": 5180
    },
    {
      "epoch": 34.02,
      "learning_rate": 7.177479492915736e-06,
      "loss": 0.058,
      "step": 5190
    },
    {
      "epoch": 34.02,
      "learning_rate": 7.084265473527218e-06,
      "loss": 0.1914,
      "step": 5200
    },
    {
      "epoch": 34.02,
      "learning_rate": 6.991051454138703e-06,
      "loss": 0.0002,
      "step": 5210
    },
    {
      "epoch": 34.02,
      "learning_rate": 6.897837434750187e-06,
      "loss": 0.0002,
      "step": 5220
    },
    {
      "epoch": 34.02,
      "learning_rate": 6.80462341536167e-06,
      "loss": 0.0536,
      "step": 5230
    },
    {
      "epoch": 34.02,
      "learning_rate": 6.7114093959731546e-06,
      "loss": 0.0026,
      "step": 5240
    },
    {
      "epoch": 34.03,
      "learning_rate": 6.618195376584639e-06,
      "loss": 0.0611,
      "step": 5250
    },
    {
      "epoch": 34.03,
      "eval_accuracy": 0.9239061509194674,
      "eval_loss": 0.46320828795433044,
      "eval_runtime": 1512.5189,
      "eval_samples_per_second": 3.128,
      "eval_steps_per_second": 0.391,
      "step": 5250
    },
    {
      "epoch": 35.0,
      "learning_rate": 6.524981357196122e-06,
      "loss": 0.0098,
      "step": 5260
    },
    {
      "epoch": 35.0,
      "learning_rate": 6.4317673378076065e-06,
      "loss": 0.1163,
      "step": 5270
    },
    {
      "epoch": 35.01,
      "learning_rate": 6.338553318419091e-06,
      "loss": 0.0005,
      "step": 5280
    },
    {
      "epoch": 35.01,
      "learning_rate": 6.245339299030575e-06,
      "loss": 0.0001,
      "step": 5290
    },
    {
      "epoch": 35.01,
      "learning_rate": 6.1521252796420584e-06,
      "loss": 0.0001,
      "step": 5300
    },
    {
      "epoch": 35.01,
      "learning_rate": 6.058911260253542e-06,
      "loss": 0.1025,
      "step": 5310
    },
    {
      "epoch": 35.01,
      "learning_rate": 5.965697240865026e-06,
      "loss": 0.0519,
      "step": 5320
    },
    {
      "epoch": 35.01,
      "learning_rate": 5.87248322147651e-06,
      "loss": 0.0053,
      "step": 5330
    },
    {
      "epoch": 35.02,
      "learning_rate": 5.779269202087995e-06,
      "loss": 0.0002,
      "step": 5340
    },
    {
      "epoch": 35.02,
      "learning_rate": 5.686055182699478e-06,
      "loss": 0.0001,
      "step": 5350
    },
    {
      "epoch": 35.02,
      "learning_rate": 5.592841163310962e-06,
      "loss": 0.0171,
      "step": 5360
    },
    {
      "epoch": 35.02,
      "learning_rate": 5.4996271439224465e-06,
      "loss": 0.0003,
      "step": 5370
    },
    {
      "epoch": 35.02,
      "learning_rate": 5.40641312453393e-06,
      "loss": 0.0002,
      "step": 5380
    },
    {
      "epoch": 35.02,
      "learning_rate": 5.313199105145414e-06,
      "loss": 0.0001,
      "step": 5390
    },
    {
      "epoch": 35.03,
      "learning_rate": 5.2199850857568985e-06,
      "loss": 0.0003,
      "step": 5400
    },
    {
      "epoch": 35.03,
      "eval_accuracy": 0.9367998309025576,
      "eval_loss": 0.37564367055892944,
      "eval_runtime": 1530.2134,
      "eval_samples_per_second": 3.092,
      "eval_steps_per_second": 0.387,
      "step": 5400
    },
    {
      "epoch": 36.0,
      "learning_rate": 5.126771066368382e-06,
      "loss": 0.0487,
      "step": 5410
    },
    {
      "epoch": 36.0,
      "learning_rate": 5.033557046979865e-06,
      "loss": 0.0702,
      "step": 5420
    },
    {
      "epoch": 36.01,
      "learning_rate": 4.94034302759135e-06,
      "loss": 0.0002,
      "step": 5430
    },
    {
      "epoch": 36.01,
      "learning_rate": 4.847129008202834e-06,
      "loss": 0.0002,
      "step": 5440
    },
    {
      "epoch": 36.01,
      "learning_rate": 4.753914988814318e-06,
      "loss": 0.1035,
      "step": 5450
    },
    {
      "epoch": 36.01,
      "learning_rate": 4.660700969425802e-06,
      "loss": 0.0001,
      "step": 5460
    },
    {
      "epoch": 36.01,
      "learning_rate": 4.567486950037286e-06,
      "loss": 0.0003,
      "step": 5470
    },
    {
      "epoch": 36.01,
      "learning_rate": 4.47427293064877e-06,
      "loss": 0.0002,
      "step": 5480
    },
    {
      "epoch": 36.02,
      "learning_rate": 4.381058911260253e-06,
      "loss": 0.0722,
      "step": 5490
    },
    {
      "epoch": 36.02,
      "learning_rate": 4.287844891871738e-06,
      "loss": 0.0002,
      "step": 5500
    },
    {
      "epoch": 36.02,
      "learning_rate": 4.194630872483222e-06,
      "loss": 0.0002,
      "step": 5510
    },
    {
      "epoch": 36.02,
      "learning_rate": 4.101416853094705e-06,
      "loss": 0.0002,
      "step": 5520
    },
    {
      "epoch": 36.02,
      "learning_rate": 4.00820283370619e-06,
      "loss": 0.0002,
      "step": 5530
    },
    {
      "epoch": 36.02,
      "learning_rate": 3.914988814317674e-06,
      "loss": 0.0001,
      "step": 5540
    },
    {
      "epoch": 36.03,
      "learning_rate": 3.821774794929157e-06,
      "loss": 0.0003,
      "step": 5550
    },
    {
      "epoch": 36.03,
      "eval_accuracy": 0.9429296131896006,
      "eval_loss": 0.37451043725013733,
      "eval_runtime": 1501.8486,
      "eval_samples_per_second": 3.15,
      "eval_steps_per_second": 0.394,
      "step": 5550
    },
    {
      "epoch": 37.0,
      "learning_rate": 3.728560775540642e-06,
      "loss": 0.0014,
      "step": 5560
    },
    {
      "epoch": 37.0,
      "learning_rate": 3.6353467561521254e-06,
      "loss": 0.0005,
      "step": 5570
    },
    {
      "epoch": 37.01,
      "learning_rate": 3.542132736763609e-06,
      "loss": 0.0004,
      "step": 5580
    },
    {
      "epoch": 37.01,
      "learning_rate": 3.4489187173750935e-06,
      "loss": 0.0001,
      "step": 5590
    },
    {
      "epoch": 37.01,
      "learning_rate": 3.3557046979865773e-06,
      "loss": 0.0001,
      "step": 5600
    },
    {
      "epoch": 37.01,
      "learning_rate": 3.262490678598061e-06,
      "loss": 0.0001,
      "step": 5610
    },
    {
      "epoch": 37.01,
      "learning_rate": 3.1692766592095454e-06,
      "loss": 0.0001,
      "step": 5620
    },
    {
      "epoch": 37.01,
      "learning_rate": 3.0760626398210292e-06,
      "loss": 0.0003,
      "step": 5630
    },
    {
      "epoch": 37.02,
      "learning_rate": 2.982848620432513e-06,
      "loss": 0.0101,
      "step": 5640
    },
    {
      "epoch": 37.02,
      "learning_rate": 2.8896346010439973e-06,
      "loss": 0.0004,
      "step": 5650
    },
    {
      "epoch": 37.02,
      "learning_rate": 2.796420581655481e-06,
      "loss": 0.0005,
      "step": 5660
    },
    {
      "epoch": 37.02,
      "learning_rate": 2.703206562266965e-06,
      "loss": 0.0311,
      "step": 5670
    },
    {
      "epoch": 37.02,
      "learning_rate": 2.6099925428784492e-06,
      "loss": 0.0001,
      "step": 5680
    },
    {
      "epoch": 37.02,
      "learning_rate": 2.5167785234899326e-06,
      "loss": 0.0037,
      "step": 5690
    },
    {
      "epoch": 37.03,
      "learning_rate": 2.423564504101417e-06,
      "loss": 0.163,
      "step": 5700
    },
    {
      "epoch": 37.03,
      "eval_accuracy": 0.9382794335235679,
      "eval_loss": 0.3966895043849945,
      "eval_runtime": 1527.4679,
      "eval_samples_per_second": 3.097,
      "eval_steps_per_second": 0.388,
      "step": 5700
    },
    {
      "epoch": 38.0,
      "learning_rate": 2.330350484712901e-06,
      "loss": 0.0331,
      "step": 5710
    },
    {
      "epoch": 38.0,
      "learning_rate": 2.237136465324385e-06,
      "loss": 0.0008,
      "step": 5720
    },
    {
      "epoch": 38.01,
      "learning_rate": 2.143922445935869e-06,
      "loss": 0.0002,
      "step": 5730
    },
    {
      "epoch": 38.01,
      "learning_rate": 2.0507084265473527e-06,
      "loss": 0.0697,
      "step": 5740
    },
    {
      "epoch": 38.01,
      "learning_rate": 1.957494407158837e-06,
      "loss": 0.0013,
      "step": 5750
    },
    {
      "epoch": 38.01,
      "learning_rate": 1.864280387770321e-06,
      "loss": 0.0001,
      "step": 5760
    },
    {
      "epoch": 38.01,
      "learning_rate": 1.7710663683818046e-06,
      "loss": 0.0001,
      "step": 5770
    },
    {
      "epoch": 38.01,
      "learning_rate": 1.6778523489932886e-06,
      "loss": 0.0001,
      "step": 5780
    },
    {
      "epoch": 38.02,
      "learning_rate": 1.5846383296047727e-06,
      "loss": 0.0008,
      "step": 5790
    },
    {
      "epoch": 38.02,
      "learning_rate": 1.4914243102162565e-06,
      "loss": 0.0001,
      "step": 5800
    },
    {
      "epoch": 38.02,
      "learning_rate": 1.3982102908277406e-06,
      "loss": 0.0001,
      "step": 5810
    },
    {
      "epoch": 38.02,
      "learning_rate": 1.3049962714392246e-06,
      "loss": 0.0001,
      "step": 5820
    },
    {
      "epoch": 38.02,
      "learning_rate": 1.2117822520507085e-06,
      "loss": 0.0447,
      "step": 5830
    },
    {
      "epoch": 38.02,
      "learning_rate": 1.1185682326621925e-06,
      "loss": 0.0003,
      "step": 5840
    },
    {
      "epoch": 38.03,
      "learning_rate": 1.0253542132736763e-06,
      "loss": 0.0059,
      "step": 5850
    },
    {
      "epoch": 38.03,
      "eval_accuracy": 0.9389135489325724,
      "eval_loss": 0.38082581758499146,
      "eval_runtime": 1532.2809,
      "eval_samples_per_second": 3.088,
      "eval_steps_per_second": 0.386,
      "step": 5850
    },
    {
      "epoch": 39.0,
      "learning_rate": 9.321401938851605e-07,
      "loss": 0.0002,
      "step": 5860
    },
    {
      "epoch": 39.0,
      "learning_rate": 8.389261744966443e-07,
      "loss": 0.0002,
      "step": 5870
    },
    {
      "epoch": 39.01,
      "learning_rate": 7.457121551081283e-07,
      "loss": 0.0002,
      "step": 5880
    },
    {
      "epoch": 39.01,
      "learning_rate": 6.524981357196123e-07,
      "loss": 0.0001,
      "step": 5890
    },
    {
      "epoch": 39.01,
      "learning_rate": 5.592841163310962e-07,
      "loss": 0.0001,
      "step": 5900
    },
    {
      "epoch": 39.01,
      "learning_rate": 4.6607009694258024e-07,
      "loss": 0.0001,
      "step": 5910
    },
    {
      "epoch": 39.01,
      "learning_rate": 3.7285607755406413e-07,
      "loss": 0.0002,
      "step": 5920
    },
    {
      "epoch": 39.01,
      "learning_rate": 2.796420581655481e-07,
      "loss": 0.0621,
      "step": 5930
    },
    {
      "epoch": 39.02,
      "learning_rate": 1.8642803877703207e-07,
      "loss": 0.0006,
      "step": 5940
    },
    {
      "epoch": 39.02,
      "learning_rate": 9.321401938851603e-08,
      "loss": 0.0002,
      "step": 5950
    },
    {
      "epoch": 39.02,
      "learning_rate": 0.0,
      "loss": 0.0003,
      "step": 5960
    },
    {
      "epoch": 39.02,
      "eval_accuracy": 0.9395476643415769,
      "eval_loss": 0.3823794424533844,
      "eval_runtime": 1509.7311,
      "eval_samples_per_second": 3.134,
      "eval_steps_per_second": 0.392,
      "step": 5960
    },
    {
      "epoch": 39.02,
      "step": 5960,
      "total_flos": 5.931516658803396e+19,
      "train_loss": 0.17179854766711056,
      "train_runtime": 92364.9889,
      "train_samples_per_second": 0.516,
      "train_steps_per_second": 0.065
    },
    {
      "epoch": 39.02,
      "eval_accuracy": 0.9614663611747553,
      "eval_loss": 0.21806761622428894,
      "eval_runtime": 1828.495,
      "eval_samples_per_second": 2.626,
      "eval_steps_per_second": 0.329,
      "step": 5960
    },
    {
      "epoch": 39.02,
      "eval_accuracy": 0.9614663611747553,
      "eval_loss": 0.21806761622428894,
      "eval_runtime": 1855.3329,
      "eval_samples_per_second": 2.588,
      "eval_steps_per_second": 0.324,
      "step": 5960
    },
    {
      "epoch": 39.02,
      "eval_accuracy": 0.9614663611747553,
      "eval_loss": 0.21806758642196655,
      "eval_runtime": 1853.612,
      "eval_samples_per_second": 2.59,
      "eval_steps_per_second": 0.324,
      "step": 5960
    }
  ],
  "logging_steps": 10,
  "max_steps": 5960,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 500,
  "total_flos": 5.931516658803396e+19,
  "trial_name": null,
  "trial_params": null
}