{
  "best_metric": 0.85,
  "best_model_checkpoint": "vit-base-patch16-224-dmae-va-U5-42B/checkpoint-209",
  "epoch": 37.935483870967744,
  "eval_steps": 500,
  "global_step": 294,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.9,
      "eval_accuracy": 0.4666666666666667,
      "eval_loss": 1.3100930452346802,
      "eval_runtime": 2.1723,
      "eval_samples_per_second": 27.62,
      "eval_steps_per_second": 0.921,
      "step": 7
    },
    {
      "epoch": 1.29,
      "grad_norm": 2.6547346115112305,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 1.408,
      "step": 10
    },
    {
      "epoch": 1.94,
      "eval_accuracy": 0.48333333333333334,
      "eval_loss": 1.1884137392044067,
      "eval_runtime": 1.6114,
      "eval_samples_per_second": 37.236,
      "eval_steps_per_second": 1.241,
      "step": 15
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.5833563804626465,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 1.1286,
      "step": 20
    },
    {
      "epoch": 2.97,
      "eval_accuracy": 0.5166666666666667,
      "eval_loss": 0.9475716352462769,
      "eval_runtime": 1.6132,
      "eval_samples_per_second": 37.192,
      "eval_steps_per_second": 1.24,
      "step": 23
    },
    {
      "epoch": 3.87,
      "grad_norm": 2.2686269283294678,
      "learning_rate": 5e-05,
      "loss": 0.7589,
      "step": 30
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.75,
      "eval_loss": 0.7636658549308777,
      "eval_runtime": 1.6545,
      "eval_samples_per_second": 36.264,
      "eval_steps_per_second": 1.209,
      "step": 31
    },
    {
      "epoch": 4.9,
      "eval_accuracy": 0.6833333333333333,
      "eval_loss": 0.7185887694358826,
      "eval_runtime": 1.6444,
      "eval_samples_per_second": 36.488,
      "eval_steps_per_second": 1.216,
      "step": 38
    },
    {
      "epoch": 5.16,
      "grad_norm": 2.46492600440979,
      "learning_rate": 4.810606060606061e-05,
      "loss": 0.4786,
      "step": 40
    },
    {
      "epoch": 5.94,
      "eval_accuracy": 0.7833333333333333,
      "eval_loss": 0.619183361530304,
      "eval_runtime": 1.6099,
      "eval_samples_per_second": 37.269,
      "eval_steps_per_second": 1.242,
      "step": 46
    },
    {
      "epoch": 6.45,
      "grad_norm": 2.310150384902954,
      "learning_rate": 4.621212121212121e-05,
      "loss": 0.2874,
      "step": 50
    },
    {
      "epoch": 6.97,
      "eval_accuracy": 0.7833333333333333,
      "eval_loss": 0.6195262670516968,
      "eval_runtime": 1.7294,
      "eval_samples_per_second": 34.693,
      "eval_steps_per_second": 1.156,
      "step": 54
    },
    {
      "epoch": 7.74,
      "grad_norm": 2.976644277572632,
      "learning_rate": 4.431818181818182e-05,
      "loss": 0.2027,
      "step": 60
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.7833333333333333,
      "eval_loss": 0.5958878993988037,
      "eval_runtime": 1.6615,
      "eval_samples_per_second": 36.113,
      "eval_steps_per_second": 1.204,
      "step": 62
    },
    {
      "epoch": 8.9,
      "eval_accuracy": 0.7666666666666667,
      "eval_loss": 0.6104174852371216,
      "eval_runtime": 2.1769,
      "eval_samples_per_second": 27.562,
      "eval_steps_per_second": 0.919,
      "step": 69
    },
    {
      "epoch": 9.03,
      "grad_norm": 1.530189871788025,
      "learning_rate": 4.242424242424243e-05,
      "loss": 0.1662,
      "step": 70
    },
    {
      "epoch": 9.94,
      "eval_accuracy": 0.75,
      "eval_loss": 0.7297105193138123,
      "eval_runtime": 1.8244,
      "eval_samples_per_second": 32.887,
      "eval_steps_per_second": 1.096,
      "step": 77
    },
    {
      "epoch": 10.32,
      "grad_norm": 2.815868377685547,
      "learning_rate": 4.053030303030303e-05,
      "loss": 0.1462,
      "step": 80
    },
    {
      "epoch": 10.97,
      "eval_accuracy": 0.7666666666666667,
      "eval_loss": 0.7851839065551758,
      "eval_runtime": 1.617,
      "eval_samples_per_second": 37.107,
      "eval_steps_per_second": 1.237,
      "step": 85
    },
    {
      "epoch": 11.61,
      "grad_norm": 3.3121185302734375,
      "learning_rate": 3.8636363636363636e-05,
      "loss": 0.1419,
      "step": 90
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.7166666666666667,
      "eval_loss": 0.8637493848800659,
      "eval_runtime": 1.7943,
      "eval_samples_per_second": 33.44,
      "eval_steps_per_second": 1.115,
      "step": 93
    },
    {
      "epoch": 12.9,
      "grad_norm": 1.8678604364395142,
      "learning_rate": 3.6742424242424246e-05,
      "loss": 0.1199,
      "step": 100
    },
    {
      "epoch": 12.9,
      "eval_accuracy": 0.7333333333333333,
      "eval_loss": 0.6797261834144592,
      "eval_runtime": 1.6329,
      "eval_samples_per_second": 36.745,
      "eval_steps_per_second": 1.225,
      "step": 100
    },
    {
      "epoch": 13.94,
      "eval_accuracy": 0.7666666666666667,
      "eval_loss": 0.766004204750061,
      "eval_runtime": 2.1818,
      "eval_samples_per_second": 27.5,
      "eval_steps_per_second": 0.917,
      "step": 108
    },
    {
      "epoch": 14.19,
      "grad_norm": 1.7959256172180176,
      "learning_rate": 3.484848484848485e-05,
      "loss": 0.0949,
      "step": 110
    },
    {
      "epoch": 14.97,
      "eval_accuracy": 0.7166666666666667,
      "eval_loss": 0.7385975122451782,
      "eval_runtime": 1.63,
      "eval_samples_per_second": 36.809,
      "eval_steps_per_second": 1.227,
      "step": 116
    },
    {
      "epoch": 15.48,
      "grad_norm": 3.5541183948516846,
      "learning_rate": 3.295454545454545e-05,
      "loss": 0.0901,
      "step": 120
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.7,
      "eval_loss": 1.0125622749328613,
      "eval_runtime": 2.1501,
      "eval_samples_per_second": 27.905,
      "eval_steps_per_second": 0.93,
      "step": 124
    },
    {
      "epoch": 16.77,
      "grad_norm": 1.4506371021270752,
      "learning_rate": 3.106060606060606e-05,
      "loss": 0.0808,
      "step": 130
    },
    {
      "epoch": 16.9,
      "eval_accuracy": 0.8,
      "eval_loss": 0.7060254812240601,
      "eval_runtime": 1.656,
      "eval_samples_per_second": 36.232,
      "eval_steps_per_second": 1.208,
      "step": 131
    },
    {
      "epoch": 17.94,
      "eval_accuracy": 0.7833333333333333,
      "eval_loss": 0.7856984734535217,
      "eval_runtime": 1.9667,
      "eval_samples_per_second": 30.508,
      "eval_steps_per_second": 1.017,
      "step": 139
    },
    {
      "epoch": 18.06,
      "grad_norm": 1.090044617652893,
      "learning_rate": 2.916666666666667e-05,
      "loss": 0.102,
      "step": 140
    },
    {
      "epoch": 18.97,
      "eval_accuracy": 0.8,
      "eval_loss": 0.741092324256897,
      "eval_runtime": 1.6707,
      "eval_samples_per_second": 35.913,
      "eval_steps_per_second": 1.197,
      "step": 147
    },
    {
      "epoch": 19.35,
      "grad_norm": 1.416364073753357,
      "learning_rate": 2.7272727272727273e-05,
      "loss": 0.0706,
      "step": 150
    },
    {
      "epoch": 20.0,
      "eval_accuracy": 0.8166666666666667,
      "eval_loss": 0.7340261340141296,
      "eval_runtime": 1.6384,
      "eval_samples_per_second": 36.622,
      "eval_steps_per_second": 1.221,
      "step": 155
    },
    {
      "epoch": 20.65,
      "grad_norm": 1.222724199295044,
      "learning_rate": 2.537878787878788e-05,
      "loss": 0.0582,
      "step": 160
    },
    {
      "epoch": 20.9,
      "eval_accuracy": 0.75,
      "eval_loss": 0.8588929772377014,
      "eval_runtime": 1.6461,
      "eval_samples_per_second": 36.45,
      "eval_steps_per_second": 1.215,
      "step": 162
    },
    {
      "epoch": 21.94,
      "grad_norm": 0.9567591547966003,
      "learning_rate": 2.3484848484848487e-05,
      "loss": 0.0687,
      "step": 170
    },
    {
      "epoch": 21.94,
      "eval_accuracy": 0.7666666666666667,
      "eval_loss": 0.8546371459960938,
      "eval_runtime": 1.6207,
      "eval_samples_per_second": 37.022,
      "eval_steps_per_second": 1.234,
      "step": 170
    },
    {
      "epoch": 22.97,
      "eval_accuracy": 0.7666666666666667,
      "eval_loss": 0.7760646939277649,
      "eval_runtime": 1.9412,
      "eval_samples_per_second": 30.909,
      "eval_steps_per_second": 1.03,
      "step": 178
    },
    {
      "epoch": 23.23,
      "grad_norm": 1.38846755027771,
      "learning_rate": 2.1590909090909093e-05,
      "loss": 0.0633,
      "step": 180
    },
    {
      "epoch": 24.0,
      "eval_accuracy": 0.7666666666666667,
      "eval_loss": 0.8112250566482544,
      "eval_runtime": 1.5795,
      "eval_samples_per_second": 37.988,
      "eval_steps_per_second": 1.266,
      "step": 186
    },
    {
      "epoch": 24.52,
      "grad_norm": 0.9209253191947937,
      "learning_rate": 1.9696969696969697e-05,
      "loss": 0.0626,
      "step": 190
    },
    {
      "epoch": 24.9,
      "eval_accuracy": 0.8333333333333334,
      "eval_loss": 0.6942941546440125,
      "eval_runtime": 1.6177,
      "eval_samples_per_second": 37.089,
      "eval_steps_per_second": 1.236,
      "step": 193
    },
    {
      "epoch": 25.81,
      "grad_norm": 0.7989098429679871,
      "learning_rate": 1.7803030303030303e-05,
      "loss": 0.0578,
      "step": 200
    },
    {
      "epoch": 25.94,
      "eval_accuracy": 0.7833333333333333,
      "eval_loss": 0.8592663407325745,
      "eval_runtime": 1.6312,
      "eval_samples_per_second": 36.783,
      "eval_steps_per_second": 1.226,
      "step": 201
    },
    {
      "epoch": 26.97,
      "eval_accuracy": 0.85,
      "eval_loss": 0.7214663028717041,
      "eval_runtime": 1.6092,
      "eval_samples_per_second": 37.286,
      "eval_steps_per_second": 1.243,
      "step": 209
    },
    {
      "epoch": 27.1,
      "grad_norm": 1.5267945528030396,
      "learning_rate": 1.590909090909091e-05,
      "loss": 0.0434,
      "step": 210
    },
    {
      "epoch": 28.0,
      "eval_accuracy": 0.8,
      "eval_loss": 0.8150014281272888,
      "eval_runtime": 1.5946,
      "eval_samples_per_second": 37.628,
      "eval_steps_per_second": 1.254,
      "step": 217
    },
    {
      "epoch": 28.39,
      "grad_norm": 1.5845617055892944,
      "learning_rate": 1.4015151515151515e-05,
      "loss": 0.0492,
      "step": 220
    },
    {
      "epoch": 28.9,
      "eval_accuracy": 0.7833333333333333,
      "eval_loss": 0.7833768725395203,
      "eval_runtime": 1.622,
      "eval_samples_per_second": 36.992,
      "eval_steps_per_second": 1.233,
      "step": 224
    },
    {
      "epoch": 29.68,
      "grad_norm": 1.6783336400985718,
      "learning_rate": 1.2121212121212122e-05,
      "loss": 0.0582,
      "step": 230
    },
    {
      "epoch": 29.94,
      "eval_accuracy": 0.7833333333333333,
      "eval_loss": 0.7843878269195557,
      "eval_runtime": 1.6057,
      "eval_samples_per_second": 37.367,
      "eval_steps_per_second": 1.246,
      "step": 232
    },
    {
      "epoch": 30.97,
      "grad_norm": 0.9550512433052063,
      "learning_rate": 1.0227272727272729e-05,
      "loss": 0.0515,
      "step": 240
    },
    {
      "epoch": 30.97,
      "eval_accuracy": 0.7666666666666667,
      "eval_loss": 0.7972704172134399,
      "eval_runtime": 1.5833,
      "eval_samples_per_second": 37.895,
      "eval_steps_per_second": 1.263,
      "step": 240
    },
    {
      "epoch": 32.0,
      "eval_accuracy": 0.8,
      "eval_loss": 0.7744101285934448,
      "eval_runtime": 1.6139,
      "eval_samples_per_second": 37.177,
      "eval_steps_per_second": 1.239,
      "step": 248
    },
    {
      "epoch": 32.26,
      "grad_norm": 1.3412911891937256,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.0487,
      "step": 250
    },
    {
      "epoch": 32.9,
      "eval_accuracy": 0.75,
      "eval_loss": 0.8613999485969543,
      "eval_runtime": 1.6004,
      "eval_samples_per_second": 37.491,
      "eval_steps_per_second": 1.25,
      "step": 255
    },
    {
      "epoch": 33.55,
      "grad_norm": 1.1709736585617065,
      "learning_rate": 6.43939393939394e-06,
      "loss": 0.0455,
      "step": 260
    },
    {
      "epoch": 33.94,
      "eval_accuracy": 0.7666666666666667,
      "eval_loss": 0.8194929957389832,
      "eval_runtime": 1.6082,
      "eval_samples_per_second": 37.309,
      "eval_steps_per_second": 1.244,
      "step": 263
    },
    {
      "epoch": 34.84,
      "grad_norm": 1.126597285270691,
      "learning_rate": 4.5454545454545455e-06,
      "loss": 0.0329,
      "step": 270
    },
    {
      "epoch": 34.97,
      "eval_accuracy": 0.7666666666666667,
      "eval_loss": 0.832683801651001,
      "eval_runtime": 2.1485,
      "eval_samples_per_second": 27.927,
      "eval_steps_per_second": 0.931,
      "step": 271
    },
    {
      "epoch": 36.0,
      "eval_accuracy": 0.7666666666666667,
      "eval_loss": 0.8889188170433044,
      "eval_runtime": 1.6066,
      "eval_samples_per_second": 37.345,
      "eval_steps_per_second": 1.245,
      "step": 279
    },
    {
      "epoch": 36.13,
      "grad_norm": 1.9044275283813477,
      "learning_rate": 2.651515151515152e-06,
      "loss": 0.0447,
      "step": 280
    },
    {
      "epoch": 36.9,
      "eval_accuracy": 0.7666666666666667,
      "eval_loss": 0.8705342411994934,
      "eval_runtime": 1.6111,
      "eval_samples_per_second": 37.243,
      "eval_steps_per_second": 1.241,
      "step": 286
    },
    {
      "epoch": 37.42,
      "grad_norm": 1.5608246326446533,
      "learning_rate": 7.575757575757576e-07,
      "loss": 0.0445,
      "step": 290
    },
    {
      "epoch": 37.94,
      "eval_accuracy": 0.7666666666666667,
      "eval_loss": 0.8694610595703125,
      "eval_runtime": 1.8543,
      "eval_samples_per_second": 32.356,
      "eval_steps_per_second": 1.079,
      "step": 294
    },
    {
      "epoch": 37.94,
      "step": 294,
      "total_flos": 2.864620236542755e+18,
      "train_loss": 0.20484183196510589,
      "train_runtime": 1635.0256,
      "train_samples_per_second": 25.02,
      "train_steps_per_second": 0.18
    }
  ],
  "logging_steps": 10,
  "max_steps": 294,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 42,
  "save_steps": 500,
  "total_flos": 2.864620236542755e+18,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}