{
  "best_metric": 0.09177211672067642,
  "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": 2.1739912033081055,
      "learning_rate": 0.00019800000000000002,
      "loss": 1.2716,
      "step": 10
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4669973850250244,
      "learning_rate": 0.000196,
      "loss": 1.0636,
      "step": 20
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3152661323547363,
      "learning_rate": 0.000194,
      "loss": 0.8686,
      "step": 30
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.30520761013031,
      "learning_rate": 0.000192,
      "loss": 0.8709,
      "step": 40
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.462522506713867,
      "learning_rate": 0.00019,
      "loss": 0.8241,
      "step": 50
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.773179531097412,
      "learning_rate": 0.000188,
      "loss": 0.743,
      "step": 60
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8740523457527161,
      "learning_rate": 0.00018600000000000002,
      "loss": 0.7111,
      "step": 70
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.179422378540039,
      "learning_rate": 0.00018400000000000003,
      "loss": 0.7327,
      "step": 80
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.8877387046813965,
      "learning_rate": 0.000182,
      "loss": 0.8438,
      "step": 90
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.021406888961792,
      "learning_rate": 0.00018,
      "loss": 0.8785,
      "step": 100
    },
    {
      "epoch": 0.4,
      "eval_accuracy": 0.711,
      "eval_loss": 0.779495358467102,
      "eval_runtime": 14.6516,
      "eval_samples_per_second": 68.252,
      "eval_steps_per_second": 8.531,
      "step": 100
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4613615274429321,
      "learning_rate": 0.00017800000000000002,
      "loss": 0.767,
      "step": 110
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6331909894943237,
      "learning_rate": 0.00017600000000000002,
      "loss": 0.9023,
      "step": 120
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2065010070800781,
      "learning_rate": 0.000174,
      "loss": 0.7347,
      "step": 130
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.609735369682312,
      "learning_rate": 0.000172,
      "loss": 0.5344,
      "step": 140
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.425642251968384,
      "learning_rate": 0.00017,
      "loss": 0.5897,
      "step": 150
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7095292210578918,
      "learning_rate": 0.000168,
      "loss": 0.6487,
      "step": 160
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6857857704162598,
      "learning_rate": 0.000166,
      "loss": 0.479,
      "step": 170
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.3737733364105225,
      "learning_rate": 0.000164,
      "loss": 0.6547,
      "step": 180
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.8827691078186035,
      "learning_rate": 0.000162,
      "loss": 0.667,
      "step": 190
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.998505711555481,
      "learning_rate": 0.00016,
      "loss": 0.7076,
      "step": 200
    },
    {
      "epoch": 0.8,
      "eval_accuracy": 0.818,
      "eval_loss": 0.5420999526977539,
      "eval_runtime": 14.8716,
      "eval_samples_per_second": 67.242,
      "eval_steps_per_second": 8.405,
      "step": 200
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6601933240890503,
      "learning_rate": 0.00015800000000000002,
      "loss": 0.5418,
      "step": 210
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.71398663520813,
      "learning_rate": 0.00015600000000000002,
      "loss": 0.6696,
      "step": 220
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.1173503398895264,
      "learning_rate": 0.000154,
      "loss": 0.5191,
      "step": 230
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7604912519454956,
      "learning_rate": 0.000152,
      "loss": 0.5596,
      "step": 240
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8370306491851807,
      "learning_rate": 0.00015000000000000001,
      "loss": 0.4941,
      "step": 250
    },
    {
      "epoch": 1.04,
      "grad_norm": 3.0072226524353027,
      "learning_rate": 0.000148,
      "loss": 0.417,
      "step": 260
    },
    {
      "epoch": 1.08,
      "grad_norm": 2.4395639896392822,
      "learning_rate": 0.000146,
      "loss": 0.5287,
      "step": 270
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.1612118482589722,
      "learning_rate": 0.000144,
      "loss": 0.4771,
      "step": 280
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5060698986053467,
      "learning_rate": 0.000142,
      "loss": 0.4211,
      "step": 290
    },
    {
      "epoch": 1.2,
      "grad_norm": 2.4940316677093506,
      "learning_rate": 0.00014,
      "loss": 0.4283,
      "step": 300
    },
    {
      "epoch": 1.2,
      "eval_accuracy": 0.876,
      "eval_loss": 0.3951060175895691,
      "eval_runtime": 14.0948,
      "eval_samples_per_second": 70.948,
      "eval_steps_per_second": 8.869,
      "step": 300
    },
    {
      "epoch": 1.24,
      "grad_norm": 3.643969774246216,
      "learning_rate": 0.000138,
      "loss": 0.4377,
      "step": 310
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.954455852508545,
      "learning_rate": 0.00013600000000000003,
      "loss": 0.4311,
      "step": 320
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4906481504440308,
      "learning_rate": 0.000134,
      "loss": 0.3393,
      "step": 330
    },
    {
      "epoch": 1.3599999999999999,
      "grad_norm": 2.0885210037231445,
      "learning_rate": 0.000132,
      "loss": 0.4909,
      "step": 340
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.9808173179626465,
      "learning_rate": 0.00013000000000000002,
      "loss": 0.4071,
      "step": 350
    },
    {
      "epoch": 1.44,
      "grad_norm": 2.143996477127075,
      "learning_rate": 0.00012800000000000002,
      "loss": 0.2979,
      "step": 360
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.164979457855225,
      "learning_rate": 0.000126,
      "loss": 0.4576,
      "step": 370
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.777462482452393,
      "learning_rate": 0.000124,
      "loss": 0.446,
      "step": 380
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1243666410446167,
      "learning_rate": 0.000122,
      "loss": 0.3795,
      "step": 390
    },
    {
      "epoch": 1.6,
      "grad_norm": 3.663288116455078,
      "learning_rate": 0.00012,
      "loss": 0.4251,
      "step": 400
    },
    {
      "epoch": 1.6,
      "eval_accuracy": 0.864,
      "eval_loss": 0.38176068663597107,
      "eval_runtime": 14.993,
      "eval_samples_per_second": 66.698,
      "eval_steps_per_second": 8.337,
      "step": 400
    },
    {
      "epoch": 1.6400000000000001,
      "grad_norm": 2.137402296066284,
      "learning_rate": 0.000118,
      "loss": 0.4443,
      "step": 410
    },
    {
      "epoch": 1.6800000000000002,
      "grad_norm": 3.767395257949829,
      "learning_rate": 0.000116,
      "loss": 0.5193,
      "step": 420
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.152035117149353,
      "learning_rate": 0.00011399999999999999,
      "loss": 0.3036,
      "step": 430
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.046035289764404,
      "learning_rate": 0.00011200000000000001,
      "loss": 0.2667,
      "step": 440
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.6602838039398193,
      "learning_rate": 0.00011000000000000002,
      "loss": 0.3035,
      "step": 450
    },
    {
      "epoch": 1.8399999999999999,
      "grad_norm": 2.508718729019165,
      "learning_rate": 0.00010800000000000001,
      "loss": 0.4932,
      "step": 460
    },
    {
      "epoch": 1.88,
      "grad_norm": 2.9392433166503906,
      "learning_rate": 0.00010600000000000002,
      "loss": 0.4191,
      "step": 470
    },
    {
      "epoch": 1.92,
      "grad_norm": 2.791663885116577,
      "learning_rate": 0.00010400000000000001,
      "loss": 0.3038,
      "step": 480
    },
    {
      "epoch": 1.96,
      "grad_norm": 2.3044326305389404,
      "learning_rate": 0.00010200000000000001,
      "loss": 0.4298,
      "step": 490
    },
    {
      "epoch": 2.0,
      "grad_norm": 3.1489250659942627,
      "learning_rate": 0.0001,
      "loss": 0.335,
      "step": 500
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.924,
      "eval_loss": 0.24736037850379944,
      "eval_runtime": 15.0667,
      "eval_samples_per_second": 66.372,
      "eval_steps_per_second": 8.296,
      "step": 500
    },
    {
      "epoch": 2.04,
      "grad_norm": 3.3036346435546875,
      "learning_rate": 9.8e-05,
      "loss": 0.1649,
      "step": 510
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.8792152404785156,
      "learning_rate": 9.6e-05,
      "loss": 0.1627,
      "step": 520
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.28887757658958435,
      "learning_rate": 9.4e-05,
      "loss": 0.3599,
      "step": 530
    },
    {
      "epoch": 2.16,
      "grad_norm": 5.17996883392334,
      "learning_rate": 9.200000000000001e-05,
      "loss": 0.3233,
      "step": 540
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.1433322429656982,
      "learning_rate": 9e-05,
      "loss": 0.222,
      "step": 550
    },
    {
      "epoch": 2.24,
      "grad_norm": 3.143852949142456,
      "learning_rate": 8.800000000000001e-05,
      "loss": 0.2713,
      "step": 560
    },
    {
      "epoch": 2.2800000000000002,
      "grad_norm": 0.4215773344039917,
      "learning_rate": 8.6e-05,
      "loss": 0.1708,
      "step": 570
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.9217822551727295,
      "learning_rate": 8.4e-05,
      "loss": 0.198,
      "step": 580
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.4554295539855957,
      "learning_rate": 8.2e-05,
      "loss": 0.1617,
      "step": 590
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2291343212127686,
      "learning_rate": 8e-05,
      "loss": 0.2286,
      "step": 600
    },
    {
      "epoch": 2.4,
      "eval_accuracy": 0.952,
      "eval_loss": 0.16752035915851593,
      "eval_runtime": 13.8882,
      "eval_samples_per_second": 72.003,
      "eval_steps_per_second": 9.0,
      "step": 600
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.3957996964454651,
      "learning_rate": 7.800000000000001e-05,
      "loss": 0.1758,
      "step": 610
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.093458414077759,
      "learning_rate": 7.6e-05,
      "loss": 0.138,
      "step": 620
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.130835771560669,
      "learning_rate": 7.4e-05,
      "loss": 0.1194,
      "step": 630
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3829611539840698,
      "learning_rate": 7.2e-05,
      "loss": 0.1629,
      "step": 640
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.17628225684165955,
      "learning_rate": 7e-05,
      "loss": 0.0954,
      "step": 650
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7156352996826172,
      "learning_rate": 6.800000000000001e-05,
      "loss": 0.172,
      "step": 660
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.950498342514038,
      "learning_rate": 6.6e-05,
      "loss": 0.1414,
      "step": 670
    },
    {
      "epoch": 2.7199999999999998,
      "grad_norm": 3.134085178375244,
      "learning_rate": 6.400000000000001e-05,
      "loss": 0.1734,
      "step": 680
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.362244129180908,
      "learning_rate": 6.2e-05,
      "loss": 0.1806,
      "step": 690
    },
    {
      "epoch": 2.8,
      "grad_norm": 5.472875118255615,
      "learning_rate": 6e-05,
      "loss": 0.1523,
      "step": 700
    },
    {
      "epoch": 2.8,
      "eval_accuracy": 0.954,
      "eval_loss": 0.1640758216381073,
      "eval_runtime": 14.3922,
      "eval_samples_per_second": 69.482,
      "eval_steps_per_second": 8.685,
      "step": 700
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.1265697479248047,
      "learning_rate": 5.8e-05,
      "loss": 0.1939,
      "step": 710
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.1513569355010986,
      "learning_rate": 5.6000000000000006e-05,
      "loss": 0.1369,
      "step": 720
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.14964208006858826,
      "learning_rate": 5.4000000000000005e-05,
      "loss": 0.1372,
      "step": 730
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.1409606784582138,
      "learning_rate": 5.2000000000000004e-05,
      "loss": 0.1503,
      "step": 740
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.1397903859615326,
      "learning_rate": 5e-05,
      "loss": 0.1655,
      "step": 750
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.6654576063156128,
      "learning_rate": 4.8e-05,
      "loss": 0.0662,
      "step": 760
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.11212094128131866,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.0362,
      "step": 770
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.8599291443824768,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.0545,
      "step": 780
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.0967254787683487,
      "learning_rate": 4.2e-05,
      "loss": 0.0428,
      "step": 790
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.8969135284423828,
      "learning_rate": 4e-05,
      "loss": 0.1346,
      "step": 800
    },
    {
      "epoch": 3.2,
      "eval_accuracy": 0.969,
      "eval_loss": 0.11203579604625702,
      "eval_runtime": 14.1593,
      "eval_samples_per_second": 70.625,
      "eval_steps_per_second": 8.828,
      "step": 800
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.32526883482933044,
      "learning_rate": 3.8e-05,
      "loss": 0.0576,
      "step": 810
    },
    {
      "epoch": 3.2800000000000002,
      "grad_norm": 3.220393419265747,
      "learning_rate": 3.6e-05,
      "loss": 0.0532,
      "step": 820
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.09663262218236923,
      "learning_rate": 3.4000000000000007e-05,
      "loss": 0.0832,
      "step": 830
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.07815458625555038,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 0.0653,
      "step": 840
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.09688442200422287,
      "learning_rate": 3e-05,
      "loss": 0.0763,
      "step": 850
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.07274580001831055,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 0.04,
      "step": 860
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.06821909546852112,
      "learning_rate": 2.6000000000000002e-05,
      "loss": 0.0778,
      "step": 870
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.8587623834609985,
      "learning_rate": 2.4e-05,
      "loss": 0.0563,
      "step": 880
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.09389644116163254,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 0.1213,
      "step": 890
    },
    {
      "epoch": 3.6,
      "grad_norm": 6.0940117835998535,
      "learning_rate": 2e-05,
      "loss": 0.0638,
      "step": 900
    },
    {
      "epoch": 3.6,
      "eval_accuracy": 0.978,
      "eval_loss": 0.10251828283071518,
      "eval_runtime": 14.4118,
      "eval_samples_per_second": 69.388,
      "eval_steps_per_second": 8.673,
      "step": 900
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.1492830514907837,
      "learning_rate": 1.8e-05,
      "loss": 0.0821,
      "step": 910
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.07156217098236084,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.023,
      "step": 920
    },
    {
      "epoch": 3.7199999999999998,
      "grad_norm": 0.07530596107244492,
      "learning_rate": 1.4000000000000001e-05,
      "loss": 0.0821,
      "step": 930
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.061225246638059616,
      "learning_rate": 1.2e-05,
      "loss": 0.0518,
      "step": 940
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.15823562443256378,
      "learning_rate": 1e-05,
      "loss": 0.0286,
      "step": 950
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.09221441298723221,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.0364,
      "step": 960
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.05617209151387215,
      "learning_rate": 6e-06,
      "loss": 0.0498,
      "step": 970
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.08119833469390869,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.049,
      "step": 980
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.08950098603963852,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.0642,
      "step": 990
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.07529381662607193,
      "learning_rate": 0.0,
      "loss": 0.0574,
      "step": 1000
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.981,
      "eval_loss": 0.09177211672067642,
      "eval_runtime": 15.0707,
      "eval_samples_per_second": 66.354,
      "eval_steps_per_second": 8.294,
      "step": 1000
    },
    {
      "epoch": 4.0,
      "step": 1000,
      "total_flos": 1.239905171570688e+18,
      "train_loss": 0.3459155881404877,
      "train_runtime": 562.8746,
      "train_samples_per_second": 28.426,
      "train_steps_per_second": 1.777
    }
  ],
  "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
}