trainer_state.json

{
  "best_metric": 0.9477333333333333,
  "best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/facebook_dino-vitb16/model_idx_0722/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": 0.21599580347537994,
      "learning_rate": 4.8108254489432715e-05,
      "loss": 0.6868,
      "step": 333
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9018666666666667,
      "eval_loss": 0.31663885712623596,
      "eval_runtime": 11.7874,
      "eval_samples_per_second": 318.137,
      "eval_steps_per_second": 5.005,
      "step": 333
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.9565061330795288,
      "learning_rate": 4.2698504093359784e-05,
      "loss": 0.1735,
      "step": 666
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9082666666666667,
      "eval_loss": 0.2975480556488037,
      "eval_runtime": 18.7028,
      "eval_samples_per_second": 200.505,
      "eval_steps_per_second": 3.155,
      "step": 666
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.4050355553627014,
      "learning_rate": 3.459431688641196e-05,
      "loss": 0.0758,
      "step": 999
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9282666666666667,
      "eval_loss": 0.26776808500289917,
      "eval_runtime": 10.979,
      "eval_samples_per_second": 341.561,
      "eval_steps_per_second": 5.374,
      "step": 999
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.25920045375823975,
      "learning_rate": 2.5029481906207065e-05,
      "loss": 0.035,
      "step": 1332
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9253333333333333,
      "eval_loss": 0.2762793004512787,
      "eval_runtime": 10.7237,
      "eval_samples_per_second": 349.692,
      "eval_steps_per_second": 5.502,
      "step": 1332
    },
    {
      "epoch": 5.0,
      "grad_norm": 2.283954620361328,
      "learning_rate": 1.5460158573036288e-05,
      "loss": 0.0107,
      "step": 1665
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9384,
      "eval_loss": 0.24382781982421875,
      "eval_runtime": 10.2782,
      "eval_samples_per_second": 364.849,
      "eval_steps_per_second": 5.74,
      "step": 1665
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.11659545451402664,
      "learning_rate": 7.343189618242854e-06,
      "loss": 0.0035,
      "step": 1998
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.9394666666666667,
      "eval_loss": 0.24542877078056335,
      "eval_runtime": 10.9591,
      "eval_samples_per_second": 342.182,
      "eval_steps_per_second": 5.384,
      "step": 1998
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.5915635228157043,
      "learning_rate": 1.91430998468726e-06,
      "loss": 0.0023,
      "step": 2331
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.9472,
      "eval_loss": 0.22822298109531403,
      "eval_runtime": 11.2204,
      "eval_samples_per_second": 334.213,
      "eval_steps_per_second": 5.258,
      "step": 2331
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.0010791812092065811,
      "learning_rate": 1.7383661915504334e-11,
      "loss": 0.0002,
      "step": 2664
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.9477333333333333,
      "eval_loss": 0.22888457775115967,
      "eval_runtime": 10.4596,
      "eval_samples_per_second": 358.521,
      "eval_steps_per_second": 5.641,
      "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
}