trainer_state.json

{
  "best_metric": 0.9229333333333334,
  "best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/facebook_dino-vitb16/model_idx_0743/checkpoints/checkpoint-2997",
  "epoch": 9.0,
  "eval_steps": 500,
  "global_step": 2997,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 2.119251012802124,
      "learning_rate": 2.9100759384327822e-05,
      "loss": 0.8626,
      "step": 333
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8658666666666667,
      "eval_loss": 0.44105789065361023,
      "eval_runtime": 11.8548,
      "eval_samples_per_second": 316.326,
      "eval_steps_per_second": 4.977,
      "step": 333
    },
    {
      "epoch": 2.0,
      "grad_norm": 23.758214950561523,
      "learning_rate": 2.6500767323004636e-05,
      "loss": 0.2162,
      "step": 666
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8914666666666666,
      "eval_loss": 0.3597257733345032,
      "eval_runtime": 12.3192,
      "eval_samples_per_second": 304.403,
      "eval_steps_per_second": 4.789,
      "step": 666
    },
    {
      "epoch": 3.0,
      "grad_norm": Infinity,
      "learning_rate": 2.2527217722455156e-05,
      "loss": 0.0874,
      "step": 999
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8898666666666667,
      "eval_loss": 0.4009188711643219,
      "eval_runtime": 11.1269,
      "eval_samples_per_second": 337.022,
      "eval_steps_per_second": 5.302,
      "step": 999
    },
    {
      "epoch": 4.0,
      "grad_norm": 2.8531486988067627,
      "learning_rate": 1.763568653571882e-05,
      "loss": 0.04,
      "step": 1332
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9,
      "eval_loss": 0.38610976934432983,
      "eval_runtime": 10.9659,
      "eval_samples_per_second": 341.97,
      "eval_steps_per_second": 5.38,
      "step": 1332
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.08183540403842926,
      "learning_rate": 1.2426252654184347e-05,
      "loss": 0.0184,
      "step": 1665
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9096,
      "eval_loss": 0.37330162525177,
      "eval_runtime": 11.0808,
      "eval_samples_per_second": 338.424,
      "eval_steps_per_second": 5.325,
      "step": 1665
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.04012862965464592,
      "learning_rate": 7.5272506870206094e-06,
      "loss": 0.008,
      "step": 1998
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.9170666666666667,
      "eval_loss": 0.35305339097976685,
      "eval_runtime": 10.9006,
      "eval_samples_per_second": 344.018,
      "eval_steps_per_second": 5.413,
      "step": 1998
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.006134804338216782,
      "learning_rate": 3.5295725730362483e-06,
      "loss": 0.003,
      "step": 2331
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.9186666666666666,
      "eval_loss": 0.3476554751396179,
      "eval_runtime": 11.0632,
      "eval_samples_per_second": 338.961,
      "eval_steps_per_second": 5.333,
      "step": 2331
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.1009151041507721,
      "learning_rate": 9.153972922231291e-07,
      "loss": 0.0013,
      "step": 2664
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.9176,
      "eval_loss": 0.34543532133102417,
      "eval_runtime": 11.0925,
      "eval_samples_per_second": 338.065,
      "eval_steps_per_second": 5.319,
      "step": 2664
    },
    {
      "epoch": 9.0,
      "grad_norm": 0.05952562391757965,
      "learning_rate": 3.2964565453985627e-11,
      "loss": 0.001,
      "step": 2997
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.9229333333333334,
      "eval_loss": 0.3378235995769501,
      "eval_runtime": 11.1041,
      "eval_samples_per_second": 337.712,
      "eval_steps_per_second": 5.313,
      "step": 2997
    }
  ],
  "logging_steps": 500,
  "max_steps": 2997,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 9,
  "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": 1.482671885603328e+19,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}