trainer_state.json

{
  "best_metric": 0.9413333333333334,
  "best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/google_vit-base-patch16-224/model_idx_0393/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": 0.26731911301612854,
      "learning_rate": 8.728616793536588e-05,
      "loss": 0.6418,
      "step": 333
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9072,
      "eval_loss": 0.3020423948764801,
      "eval_runtime": 4.2883,
      "eval_samples_per_second": 874.467,
      "eval_steps_per_second": 13.758,
      "step": 333
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.34079059958457947,
      "learning_rate": 7.947199994035401e-05,
      "loss": 0.1516,
      "step": 666
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9253333333333333,
      "eval_loss": 0.2619155943393707,
      "eval_runtime": 4.2499,
      "eval_samples_per_second": 882.373,
      "eval_steps_per_second": 13.883,
      "step": 666
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.04624754562973976,
      "learning_rate": 6.75e-05,
      "loss": 0.0795,
      "step": 999
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9250666666666667,
      "eval_loss": 0.2744958996772766,
      "eval_runtime": 4.8603,
      "eval_samples_per_second": 771.549,
      "eval_steps_per_second": 12.139,
      "step": 999
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.06554731726646423,
      "learning_rate": 5.2814167995011876e-05,
      "loss": 0.0422,
      "step": 1332
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9296,
      "eval_loss": 0.2568977177143097,
      "eval_runtime": 4.3729,
      "eval_samples_per_second": 857.563,
      "eval_steps_per_second": 13.492,
      "step": 1332
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.0884971097111702,
      "learning_rate": 3.7185832004988136e-05,
      "loss": 0.0192,
      "step": 1665
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.94,
      "eval_loss": 0.25329604744911194,
      "eval_runtime": 4.5663,
      "eval_samples_per_second": 821.231,
      "eval_steps_per_second": 12.921,
      "step": 1665
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.012672766111791134,
      "learning_rate": 2.250000000000001e-05,
      "loss": 0.0074,
      "step": 1998
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.936,
      "eval_loss": 0.2652117609977722,
      "eval_runtime": 4.5265,
      "eval_samples_per_second": 828.455,
      "eval_steps_per_second": 13.034,
      "step": 1998
    },
    {
      "epoch": 7.0,
      "grad_norm": NaN,
      "learning_rate": 1.055833996679956e-05,
      "loss": 0.0038,
      "step": 2331
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.9386666666666666,
      "eval_loss": 0.26197686791419983,
      "eval_runtime": 4.3227,
      "eval_samples_per_second": 867.517,
      "eval_steps_per_second": 13.649,
      "step": 2331
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.005029891151934862,
      "learning_rate": 2.7299887471028354e-06,
      "loss": 0.0019,
      "step": 2664
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.9408,
      "eval_loss": 0.2470119148492813,
      "eval_runtime": 4.2471,
      "eval_samples_per_second": 882.95,
      "eval_steps_per_second": 13.892,
      "step": 2664
    },
    {
      "epoch": 9.0,
      "grad_norm": 0.03651440888643265,
      "learning_rate": 2.472343088255613e-11,
      "loss": 0.0014,
      "step": 2997
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.9413333333333334,
      "eval_loss": 0.239932119846344,
      "eval_runtime": 6.1673,
      "eval_samples_per_second": 608.05,
      "eval_steps_per_second": 9.567,
      "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
}