{
  "best_metric": 0.34654033184051514,
  "best_model_checkpoint": "./vit-base-images/checkpoint-1000",
  "epoch": 4.0,
  "eval_steps": 100,
  "global_step": 1000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.04,
      "grad_norm": 1.5996569395065308,
      "learning_rate": 0.00019800000000000002,
      "loss": 1.0932,
      "step": 10
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9198634624481201,
      "learning_rate": 0.000196,
      "loss": 0.9522,
      "step": 20
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2456237077713013,
      "learning_rate": 0.000194,
      "loss": 0.6875,
      "step": 30
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7687036991119385,
      "learning_rate": 0.000192,
      "loss": 0.9009,
      "step": 40
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.880617141723633,
      "learning_rate": 0.00019,
      "loss": 0.7155,
      "step": 50
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5424929857254028,
      "learning_rate": 0.000188,
      "loss": 0.8144,
      "step": 60
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.816297173500061,
      "learning_rate": 0.00018600000000000002,
      "loss": 0.6641,
      "step": 70
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0054569244384766,
      "learning_rate": 0.00018400000000000003,
      "loss": 0.5917,
      "step": 80
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.2372283935546875,
      "learning_rate": 0.000182,
      "loss": 0.7852,
      "step": 90
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.542130708694458,
      "learning_rate": 0.00018,
      "loss": 0.7334,
      "step": 100
    },
    {
      "epoch": 0.4,
      "eval_accuracy": 0.779,
      "eval_loss": 0.6142178773880005,
      "eval_runtime": 18.2795,
      "eval_samples_per_second": 54.706,
      "eval_steps_per_second": 6.838,
      "step": 100
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4493112564086914,
      "learning_rate": 0.00017800000000000002,
      "loss": 0.5289,
      "step": 110
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.905771017074585,
      "learning_rate": 0.00017600000000000002,
      "loss": 0.6191,
      "step": 120
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.2236440181732178,
      "learning_rate": 0.000174,
      "loss": 0.5111,
      "step": 130
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.113398551940918,
      "learning_rate": 0.000172,
      "loss": 0.6606,
      "step": 140
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.4624953269958496,
      "learning_rate": 0.00017,
      "loss": 0.5002,
      "step": 150
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.324570417404175,
      "learning_rate": 0.000168,
      "loss": 0.9353,
      "step": 160
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.384814262390137,
      "learning_rate": 0.000166,
      "loss": 0.6604,
      "step": 170
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8541224598884583,
      "learning_rate": 0.000164,
      "loss": 0.4894,
      "step": 180
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.017305612564087,
      "learning_rate": 0.000162,
      "loss": 0.6219,
      "step": 190
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9483362436294556,
      "learning_rate": 0.00016,
      "loss": 0.6032,
      "step": 200
    },
    {
      "epoch": 0.8,
      "eval_accuracy": 0.808,
      "eval_loss": 0.5516341328620911,
      "eval_runtime": 14.5864,
      "eval_samples_per_second": 68.557,
      "eval_steps_per_second": 8.57,
      "step": 200
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.7376227378845215,
      "learning_rate": 0.00015800000000000002,
      "loss": 0.4968,
      "step": 210
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.563944697380066,
      "learning_rate": 0.00015600000000000002,
      "loss": 0.4505,
      "step": 220
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3606369495391846,
      "learning_rate": 0.000154,
      "loss": 0.5368,
      "step": 230
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3428421020507812,
      "learning_rate": 0.000152,
      "loss": 0.4932,
      "step": 240
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.9562724828720093,
      "learning_rate": 0.00015000000000000001,
      "loss": 0.4884,
      "step": 250
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.947496771812439,
      "learning_rate": 0.000148,
      "loss": 0.381,
      "step": 260
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.6039777994155884,
      "learning_rate": 0.000146,
      "loss": 0.6633,
      "step": 270
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.8116464614868164,
      "learning_rate": 0.000144,
      "loss": 0.3728,
      "step": 280
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6644967794418335,
      "learning_rate": 0.000142,
      "loss": 0.3299,
      "step": 290
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.5359082221984863,
      "learning_rate": 0.00014,
      "loss": 0.4725,
      "step": 300
    },
    {
      "epoch": 1.2,
      "eval_accuracy": 0.854,
      "eval_loss": 0.43897509574890137,
      "eval_runtime": 14.256,
      "eval_samples_per_second": 70.146,
      "eval_steps_per_second": 8.768,
      "step": 300
    },
    {
      "epoch": 1.24,
      "grad_norm": 2.018160581588745,
      "learning_rate": 0.000138,
      "loss": 0.3064,
      "step": 310
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.5475637912750244,
      "learning_rate": 0.00013600000000000003,
      "loss": 0.2928,
      "step": 320
    },
    {
      "epoch": 1.32,
      "grad_norm": 2.780301809310913,
      "learning_rate": 0.000134,
      "loss": 0.2959,
      "step": 330
    },
    {
      "epoch": 1.3599999999999999,
      "grad_norm": 1.0915693044662476,
      "learning_rate": 0.000132,
      "loss": 0.3152,
      "step": 340
    },
    {
      "epoch": 1.4,
      "grad_norm": 2.1470773220062256,
      "learning_rate": 0.00013000000000000002,
      "loss": 0.4123,
      "step": 350
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.054312705993652,
      "learning_rate": 0.00012800000000000002,
      "loss": 0.5676,
      "step": 360
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.8798813819885254,
      "learning_rate": 0.000126,
      "loss": 0.3909,
      "step": 370
    },
    {
      "epoch": 1.52,
      "grad_norm": 2.3789453506469727,
      "learning_rate": 0.000124,
      "loss": 0.419,
      "step": 380
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7660586833953857,
      "learning_rate": 0.000122,
      "loss": 0.516,
      "step": 390
    },
    {
      "epoch": 1.6,
      "grad_norm": 3.304502010345459,
      "learning_rate": 0.00012,
      "loss": 0.3638,
      "step": 400
    },
    {
      "epoch": 1.6,
      "eval_accuracy": 0.822,
      "eval_loss": 0.4622470438480377,
      "eval_runtime": 14.2766,
      "eval_samples_per_second": 70.045,
      "eval_steps_per_second": 8.756,
      "step": 400
    },
    {
      "epoch": 1.6400000000000001,
      "grad_norm": 3.906277656555176,
      "learning_rate": 0.000118,
      "loss": 0.3608,
      "step": 410
    },
    {
      "epoch": 1.6800000000000002,
      "grad_norm": 2.591684103012085,
      "learning_rate": 0.000116,
      "loss": 0.4414,
      "step": 420
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.6823468804359436,
      "learning_rate": 0.00011399999999999999,
      "loss": 0.3937,
      "step": 430
    },
    {
      "epoch": 1.76,
      "grad_norm": 2.4249002933502197,
      "learning_rate": 0.00011200000000000001,
      "loss": 0.2984,
      "step": 440
    },
    {
      "epoch": 1.8,
      "grad_norm": 2.575287103652954,
      "learning_rate": 0.00011000000000000002,
      "loss": 0.4073,
      "step": 450
    },
    {
      "epoch": 1.8399999999999999,
      "grad_norm": 0.8557507395744324,
      "learning_rate": 0.00010800000000000001,
      "loss": 0.3573,
      "step": 460
    },
    {
      "epoch": 1.88,
      "grad_norm": 3.4324100017547607,
      "learning_rate": 0.00010600000000000002,
      "loss": 0.3758,
      "step": 470
    },
    {
      "epoch": 1.92,
      "grad_norm": 2.3825552463531494,
      "learning_rate": 0.00010400000000000001,
      "loss": 0.2375,
      "step": 480
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.9951996207237244,
      "learning_rate": 0.00010200000000000001,
      "loss": 0.2496,
      "step": 490
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.2203187942504883,
      "learning_rate": 0.0001,
      "loss": 0.3279,
      "step": 500
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.876,
      "eval_loss": 0.3772076666355133,
      "eval_runtime": 14.2674,
      "eval_samples_per_second": 70.09,
      "eval_steps_per_second": 8.761,
      "step": 500
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.8857389688491821,
      "learning_rate": 9.8e-05,
      "loss": 0.1633,
      "step": 510
    },
    {
      "epoch": 2.08,
      "grad_norm": 5.698770046234131,
      "learning_rate": 9.6e-05,
      "loss": 0.2812,
      "step": 520
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.7683120965957642,
      "learning_rate": 9.4e-05,
      "loss": 0.1895,
      "step": 530
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.6420239806175232,
      "learning_rate": 9.200000000000001e-05,
      "loss": 0.1732,
      "step": 540
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.8955737948417664,
      "learning_rate": 9e-05,
      "loss": 0.1557,
      "step": 550
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.202012300491333,
      "learning_rate": 8.800000000000001e-05,
      "loss": 0.2851,
      "step": 560
    },
    {
      "epoch": 2.2800000000000002,
      "grad_norm": 3.6105308532714844,
      "learning_rate": 8.6e-05,
      "loss": 0.1645,
      "step": 570
    },
    {
      "epoch": 2.32,
      "grad_norm": 3.514596462249756,
      "learning_rate": 8.4e-05,
      "loss": 0.1399,
      "step": 580
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.36515474319458,
      "learning_rate": 8.2e-05,
      "loss": 0.2495,
      "step": 590
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.10514427721500397,
      "learning_rate": 8e-05,
      "loss": 0.1337,
      "step": 600
    },
    {
      "epoch": 2.4,
      "eval_accuracy": 0.869,
      "eval_loss": 0.45184341073036194,
      "eval_runtime": 14.683,
      "eval_samples_per_second": 68.106,
      "eval_steps_per_second": 8.513,
      "step": 600
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.140317440032959,
      "learning_rate": 7.800000000000001e-05,
      "loss": 0.1493,
      "step": 610
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.3709057569503784,
      "learning_rate": 7.6e-05,
      "loss": 0.164,
      "step": 620
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.097055196762085,
      "learning_rate": 7.4e-05,
      "loss": 0.208,
      "step": 630
    },
    {
      "epoch": 2.56,
      "grad_norm": 3.960178852081299,
      "learning_rate": 7.2e-05,
      "loss": 0.2337,
      "step": 640
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.339881420135498,
      "learning_rate": 7e-05,
      "loss": 0.167,
      "step": 650
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.97763729095459,
      "learning_rate": 6.800000000000001e-05,
      "loss": 0.2026,
      "step": 660
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.5411188006401062,
      "learning_rate": 6.6e-05,
      "loss": 0.0895,
      "step": 670
    },
    {
      "epoch": 2.7199999999999998,
      "grad_norm": 0.25824812054634094,
      "learning_rate": 6.400000000000001e-05,
      "loss": 0.0933,
      "step": 680
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.2557239830493927,
      "learning_rate": 6.2e-05,
      "loss": 0.1155,
      "step": 690
    },
    {
      "epoch": 2.8,
      "grad_norm": 5.947152137756348,
      "learning_rate": 6e-05,
      "loss": 0.236,
      "step": 700
    },
    {
      "epoch": 2.8,
      "eval_accuracy": 0.878,
      "eval_loss": 0.37660717964172363,
      "eval_runtime": 14.2296,
      "eval_samples_per_second": 70.276,
      "eval_steps_per_second": 8.785,
      "step": 700
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.6783535480499268,
      "learning_rate": 5.8e-05,
      "loss": 0.1437,
      "step": 710
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.7082568407058716,
      "learning_rate": 5.6000000000000006e-05,
      "loss": 0.1498,
      "step": 720
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.3654639720916748,
      "learning_rate": 5.4000000000000005e-05,
      "loss": 0.1544,
      "step": 730
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.7878735065460205,
      "learning_rate": 5.2000000000000004e-05,
      "loss": 0.2657,
      "step": 740
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.31339430809021,
      "learning_rate": 5e-05,
      "loss": 0.103,
      "step": 750
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.41359299421310425,
      "learning_rate": 4.8e-05,
      "loss": 0.0904,
      "step": 760
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.11081337183713913,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.0475,
      "step": 770
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.6292364001274109,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.0613,
      "step": 780
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.06634623557329178,
      "learning_rate": 4.2e-05,
      "loss": 0.0419,
      "step": 790
    },
    {
      "epoch": 3.2,
      "grad_norm": 3.720346212387085,
      "learning_rate": 4e-05,
      "loss": 0.0275,
      "step": 800
    },
    {
      "epoch": 3.2,
      "eval_accuracy": 0.891,
      "eval_loss": 0.3517528176307678,
      "eval_runtime": 14.2729,
      "eval_samples_per_second": 70.063,
      "eval_steps_per_second": 8.758,
      "step": 800
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.15002816915512085,
      "learning_rate": 3.8e-05,
      "loss": 0.0425,
      "step": 810
    },
    {
      "epoch": 3.2800000000000002,
      "grad_norm": 0.08299177885055542,
      "learning_rate": 3.6e-05,
      "loss": 0.0465,
      "step": 820
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.41334620118141174,
      "learning_rate": 3.4000000000000007e-05,
      "loss": 0.0434,
      "step": 830
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.5403936505317688,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 0.0301,
      "step": 840
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.08261027932167053,
      "learning_rate": 3e-05,
      "loss": 0.072,
      "step": 850
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.0293442010879517,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 0.082,
      "step": 860
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.7797234058380127,
      "learning_rate": 2.6000000000000002e-05,
      "loss": 0.0748,
      "step": 870
    },
    {
      "epoch": 3.52,
      "grad_norm": 3.523738145828247,
      "learning_rate": 2.4e-05,
      "loss": 0.1751,
      "step": 880
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.06309465318918228,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 0.0383,
      "step": 890
    },
    {
      "epoch": 3.6,
      "grad_norm": 2.1426751613616943,
      "learning_rate": 2e-05,
      "loss": 0.0427,
      "step": 900
    },
    {
      "epoch": 3.6,
      "eval_accuracy": 0.896,
      "eval_loss": 0.3709311783313751,
      "eval_runtime": 14.359,
      "eval_samples_per_second": 69.643,
      "eval_steps_per_second": 8.705,
      "step": 900
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.3229968547821045,
      "learning_rate": 1.8e-05,
      "loss": 0.0352,
      "step": 910
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.08263090997934341,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.0192,
      "step": 920
    },
    {
      "epoch": 3.7199999999999998,
      "grad_norm": 0.1414523720741272,
      "learning_rate": 1.4000000000000001e-05,
      "loss": 0.0724,
      "step": 930
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.05866268649697304,
      "learning_rate": 1.2e-05,
      "loss": 0.0289,
      "step": 940
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.08174656331539154,
      "learning_rate": 1e-05,
      "loss": 0.0264,
      "step": 950
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.07566811144351959,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.0225,
      "step": 960
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.06544584035873413,
      "learning_rate": 6e-06,
      "loss": 0.0488,
      "step": 970
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.2268047034740448,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.0423,
      "step": 980
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.05503053963184357,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.0506,
      "step": 990
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.3238757252693176,
      "learning_rate": 0.0,
      "loss": 0.0363,
      "step": 1000
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.905,
      "eval_loss": 0.34654033184051514,
      "eval_runtime": 14.8487,
      "eval_samples_per_second": 67.346,
      "eval_steps_per_second": 8.418,
      "step": 1000
    },
    {
      "epoch": 4.0,
      "step": 1000,
      "total_flos": 1.239905171570688e+18,
      "train_loss": 0.3179253642559052,
      "train_runtime": 607.1163,
      "train_samples_per_second": 26.354,
      "train_steps_per_second": 1.647
    }
  ],
  "logging_steps": 10,
  "max_steps": 1000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.239905171570688e+18,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}