trainer_state.json

{
  "best_metric": 0.9445333333333333,
  "best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/google_vit-base-patch16-224/model_idx_0416/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": 13.102588653564453,
      "learning_rate": 9.698463103929542e-05,
      "loss": 0.6059,
      "step": 333
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.908,
      "eval_loss": 0.3004409074783325,
      "eval_runtime": 6.0886,
      "eval_samples_per_second": 615.9,
      "eval_steps_per_second": 9.69,
      "step": 333
    },
    {
      "epoch": 2.0,
      "grad_norm": 16.62833595275879,
      "learning_rate": 8.83022221559489e-05,
      "loss": 0.1328,
      "step": 666
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9216,
      "eval_loss": 0.24976040422916412,
      "eval_runtime": 5.3793,
      "eval_samples_per_second": 697.119,
      "eval_steps_per_second": 10.968,
      "step": 666
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.008633903227746487,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.0598,
      "step": 999
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9274666666666667,
      "eval_loss": 0.25252193212509155,
      "eval_runtime": 5.4728,
      "eval_samples_per_second": 685.202,
      "eval_steps_per_second": 10.781,
      "step": 999
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.045309118926525116,
      "learning_rate": 5.868240888334653e-05,
      "loss": 0.0262,
      "step": 1332
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9338666666666666,
      "eval_loss": 0.24994061887264252,
      "eval_runtime": 6.135,
      "eval_samples_per_second": 611.245,
      "eval_steps_per_second": 9.617,
      "step": 1332
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.03200969845056534,
      "learning_rate": 4.131759111665349e-05,
      "loss": 0.0113,
      "step": 1665
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9373333333333334,
      "eval_loss": 0.24572905898094177,
      "eval_runtime": 5.9252,
      "eval_samples_per_second": 632.894,
      "eval_steps_per_second": 9.958,
      "step": 1665
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.013315425254404545,
      "learning_rate": 2.500000000000001e-05,
      "loss": 0.0025,
      "step": 1998
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.944,
      "eval_loss": 0.24011430144309998,
      "eval_runtime": 5.2433,
      "eval_samples_per_second": 715.195,
      "eval_steps_per_second": 11.252,
      "step": 1998
    },
    {
      "epoch": 7.0,
      "grad_norm": NaN,
      "learning_rate": 1.1731488851999511e-05,
      "loss": 0.0015,
      "step": 2331
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.944,
      "eval_loss": 0.2434437870979309,
      "eval_runtime": 5.2703,
      "eval_samples_per_second": 711.538,
      "eval_steps_per_second": 11.195,
      "step": 2331
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.014546016231179237,
      "learning_rate": 3.033320830114261e-06,
      "loss": 0.001,
      "step": 2664
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.9442666666666667,
      "eval_loss": 0.2447790652513504,
      "eval_runtime": 5.9913,
      "eval_samples_per_second": 625.909,
      "eval_steps_per_second": 9.848,
      "step": 2664
    },
    {
      "epoch": 9.0,
      "grad_norm": 0.16623549163341522,
      "learning_rate": 2.7470478758395702e-11,
      "loss": 0.001,
      "step": 2997
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.9445333333333333,
      "eval_loss": 0.24500960111618042,
      "eval_runtime": 5.327,
      "eval_samples_per_second": 703.962,
      "eval_steps_per_second": 11.076,
      "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
}