trainer_state.json

{
  "best_metric": 0.9301333333333334,
  "best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/google_vit-base-patch16-224/model_idx_0266/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": 0.6153425574302673,
      "learning_rate": 9.619397662556435e-05,
      "loss": 0.6769,
      "step": 333
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8890666666666667,
      "eval_loss": 0.3681430518627167,
      "eval_runtime": 6.3538,
      "eval_samples_per_second": 590.202,
      "eval_steps_per_second": 9.286,
      "step": 333
    },
    {
      "epoch": 2.0,
      "grad_norm": NaN,
      "learning_rate": 8.539700818671957e-05,
      "loss": 0.1893,
      "step": 666
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9045333333333333,
      "eval_loss": 0.32579904794692993,
      "eval_runtime": 6.8117,
      "eval_samples_per_second": 550.522,
      "eval_steps_per_second": 8.662,
      "step": 666
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.08176368474960327,
      "learning_rate": 6.918863377282392e-05,
      "loss": 0.0973,
      "step": 999
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9141333333333334,
      "eval_loss": 0.29888078570365906,
      "eval_runtime": 6.6125,
      "eval_samples_per_second": 567.111,
      "eval_steps_per_second": 8.923,
      "step": 999
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.01864379644393921,
      "learning_rate": 5.005896381241413e-05,
      "loss": 0.0424,
      "step": 1332
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9090666666666667,
      "eval_loss": 0.33191385865211487,
      "eval_runtime": 5.9297,
      "eval_samples_per_second": 632.414,
      "eval_steps_per_second": 9.95,
      "step": 1332
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.03379053995013237,
      "learning_rate": 3.0920317146072576e-05,
      "loss": 0.0177,
      "step": 1665
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9226666666666666,
      "eval_loss": 0.3258868455886841,
      "eval_runtime": 5.8604,
      "eval_samples_per_second": 639.89,
      "eval_steps_per_second": 10.068,
      "step": 1665
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.022312531247735023,
      "learning_rate": 1.4686379236485708e-05,
      "loss": 0.0071,
      "step": 1998
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.9277333333333333,
      "eval_loss": 0.30410805344581604,
      "eval_runtime": 5.9144,
      "eval_samples_per_second": 634.044,
      "eval_steps_per_second": 9.976,
      "step": 1998
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.03516281023621559,
      "learning_rate": 3.82861996937452e-06,
      "loss": 0.0036,
      "step": 2331
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.9256,
      "eval_loss": 0.30599668622016907,
      "eval_runtime": 6.1594,
      "eval_samples_per_second": 608.827,
      "eval_steps_per_second": 9.579,
      "step": 2331
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.05425269156694412,
      "learning_rate": 3.476732383100867e-11,
      "loss": 0.0027,
      "step": 2664
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.9301333333333334,
      "eval_loss": 0.29848310351371765,
      "eval_runtime": 6.2427,
      "eval_samples_per_second": 600.699,
      "eval_steps_per_second": 9.451,
      "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
}