trainer_state.json

{
  "best_metric": 0.9301333333333334,
  "best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/facebook_dino-vitb16/model_idx_0068/checkpoints/checkpoint-1332",
  "epoch": 4.0,
  "eval_steps": 500,
  "global_step": 1332,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 0.1633688509464264,
      "learning_rate": 4.271931404387096e-05,
      "loss": 0.7348,
      "step": 333
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8749333333333333,
      "eval_loss": 0.39886921644210815,
      "eval_runtime": 7.1197,
      "eval_samples_per_second": 526.707,
      "eval_steps_per_second": 8.287,
      "step": 333
    },
    {
      "epoch": 2.0,
      "grad_norm": NaN,
      "learning_rate": 2.511792721482581e-05,
      "loss": 0.1888,
      "step": 666
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8989333333333334,
      "eval_loss": 0.34402725100517273,
      "eval_runtime": 7.3504,
      "eval_samples_per_second": 510.174,
      "eval_steps_per_second": 8.027,
      "step": 666
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.1593882441520691,
      "learning_rate": 7.405914276967749e-06,
      "loss": 0.0491,
      "step": 999
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9261333333333334,
      "eval_loss": 0.25977903604507446,
      "eval_runtime": 6.7881,
      "eval_samples_per_second": 552.436,
      "eval_steps_per_second": 8.692,
      "step": 999
    },
    {
      "epoch": 4.0,
      "grad_norm": 12.736542701721191,
      "learning_rate": 2.7813810714871767e-10,
      "loss": 0.0097,
      "step": 1332
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9301333333333334,
      "eval_loss": 0.23093801736831665,
      "eval_runtime": 6.9457,
      "eval_samples_per_second": 539.901,
      "eval_steps_per_second": 8.494,
      "step": 1332
    }
  ],
  "logging_steps": 500,
  "max_steps": 1332,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "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": 6.58965282490368e+18,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}