trainer_state.json

{
  "best_metric": 0.9221333333333334,
  "best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/google_vit-base-patch16-224/model_idx_0972/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": 1.178208827972412,
      "learning_rate": 0.0004809698831278217,
      "loss": 0.7783,
      "step": 333
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8250666666666666,
      "eval_loss": 0.6533358097076416,
      "eval_runtime": 4.212,
      "eval_samples_per_second": 890.318,
      "eval_steps_per_second": 14.008,
      "step": 333
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.04182523116469383,
      "learning_rate": 0.00042677669529663686,
      "loss": 0.4177,
      "step": 666
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8341333333333333,
      "eval_loss": 0.577990710735321,
      "eval_runtime": 4.1407,
      "eval_samples_per_second": 905.646,
      "eval_steps_per_second": 14.249,
      "step": 666
    },
    {
      "epoch": 3.0,
      "grad_norm": 8.366318702697754,
      "learning_rate": 0.0003456708580912725,
      "loss": 0.2604,
      "step": 999
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8645333333333334,
      "eval_loss": 0.4752730429172516,
      "eval_runtime": 4.2028,
      "eval_samples_per_second": 892.268,
      "eval_steps_per_second": 14.038,
      "step": 999
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.02503865957260132,
      "learning_rate": 0.00025,
      "loss": 0.1426,
      "step": 1332
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8693333333333333,
      "eval_loss": 0.47405683994293213,
      "eval_runtime": 4.1128,
      "eval_samples_per_second": 911.785,
      "eval_steps_per_second": 14.345,
      "step": 1332
    },
    {
      "epoch": 5.0,
      "grad_norm": 14.048666954040527,
      "learning_rate": 0.00015432914190872756,
      "loss": 0.0592,
      "step": 1665
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9026666666666666,
      "eval_loss": 0.3919893801212311,
      "eval_runtime": 4.0914,
      "eval_samples_per_second": 916.548,
      "eval_steps_per_second": 14.42,
      "step": 1665
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.10495452582836151,
      "learning_rate": 7.322330470336314e-05,
      "loss": 0.0131,
      "step": 1998
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.9154666666666667,
      "eval_loss": 0.33725982904434204,
      "eval_runtime": 4.212,
      "eval_samples_per_second": 890.32,
      "eval_steps_per_second": 14.008,
      "step": 1998
    },
    {
      "epoch": 7.0,
      "grad_norm": NaN,
      "learning_rate": 1.9143099846872597e-05,
      "loss": 0.0036,
      "step": 2331
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.9205333333333333,
      "eval_loss": 0.32280465960502625,
      "eval_runtime": 4.1314,
      "eval_samples_per_second": 907.688,
      "eval_steps_per_second": 14.281,
      "step": 2331
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.01766633242368698,
      "learning_rate": 1.7383661915504333e-10,
      "loss": 0.0015,
      "step": 2664
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.9221333333333334,
      "eval_loss": 0.3274519443511963,
      "eval_runtime": 4.2918,
      "eval_samples_per_second": 873.756,
      "eval_steps_per_second": 13.747,
      "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
}