trainer_state.json

{
  "best_metric": 0.9565333333333333,
  "best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/google_vit-base-patch16-224/model_idx_0433/checkpoints/checkpoint-2664",
  "epoch": 8.0,
  "eval_steps": 500,
  "global_step": 2664,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": Infinity,
      "learning_rate": 8.659485808097889e-05,
      "loss": 0.6068,
      "step": 333
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9394666666666667,
      "eval_loss": 0.20972700417041779,
      "eval_runtime": 11.283,
      "eval_samples_per_second": 332.357,
      "eval_steps_per_second": 5.229,
      "step": 333
    },
    {
      "epoch": 2.0,
      "grad_norm": 15.695802688598633,
      "learning_rate": 7.685730736804761e-05,
      "loss": 0.1392,
      "step": 666
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9424,
      "eval_loss": 0.19553451240062714,
      "eval_runtime": 11.1711,
      "eval_samples_per_second": 335.686,
      "eval_steps_per_second": 5.281,
      "step": 666
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.06375709921121597,
      "learning_rate": 6.226977039554153e-05,
      "loss": 0.0673,
      "step": 999
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9421333333333334,
      "eval_loss": 0.19762229919433594,
      "eval_runtime": 11.5797,
      "eval_samples_per_second": 323.842,
      "eval_steps_per_second": 5.095,
      "step": 999
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.018545817583799362,
      "learning_rate": 4.505306743117272e-05,
      "loss": 0.0279,
      "step": 1332
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9490666666666666,
      "eval_loss": 0.1943870633840561,
      "eval_runtime": 11.3923,
      "eval_samples_per_second": 329.17,
      "eval_steps_per_second": 5.179,
      "step": 1332
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.015889830887317657,
      "learning_rate": 2.7828285431465317e-05,
      "loss": 0.0113,
      "step": 1665
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9504,
      "eval_loss": 0.1930665373802185,
      "eval_runtime": 11.7198,
      "eval_samples_per_second": 319.972,
      "eval_steps_per_second": 5.034,
      "step": 1665
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.011744930408895016,
      "learning_rate": 1.3217741312837138e-05,
      "loss": 0.0043,
      "step": 1998
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.9514666666666667,
      "eval_loss": 0.1758803129196167,
      "eval_runtime": 11.1602,
      "eval_samples_per_second": 336.015,
      "eval_steps_per_second": 5.287,
      "step": 1998
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.009729577228426933,
      "learning_rate": 3.4457579724370674e-06,
      "loss": 0.0025,
      "step": 2331
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.9533333333333334,
      "eval_loss": 0.17675605416297913,
      "eval_runtime": 11.0728,
      "eval_samples_per_second": 338.667,
      "eval_steps_per_second": 5.328,
      "step": 2331
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.01093159057199955,
      "learning_rate": 3.12905914479078e-11,
      "loss": 0.0019,
      "step": 2664
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.9565333333333333,
      "eval_loss": 0.1783367395401001,
      "eval_runtime": 10.7807,
      "eval_samples_per_second": 347.845,
      "eval_steps_per_second": 5.473,
      "step": 2664
    }
  ],
  "logging_steps": 500,
  "max_steps": 2664,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 8,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.317930564980736e+19,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}