trainer_state.json

{
  "best_metric": 0.9048,
  "best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/facebook_vit-mae-base/model_idx_0104/checkpoints/checkpoint-1332",
  "epoch": 4.0,
  "eval_steps": 500,
  "global_step": 1332,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 7.073408126831055,
      "learning_rate": 7.681980515339464e-05,
      "loss": 1.5265,
      "step": 333
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8298666666666666,
      "eval_loss": 0.6374258995056152,
      "eval_runtime": 6.4005,
      "eval_samples_per_second": 585.893,
      "eval_steps_per_second": 9.218,
      "step": 333
    },
    {
      "epoch": 2.0,
      "grad_norm": Infinity,
      "learning_rate": 4.521226898668646e-05,
      "loss": 0.3828,
      "step": 666
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8741333333333333,
      "eval_loss": 0.44095417857170105,
      "eval_runtime": 6.7108,
      "eval_samples_per_second": 558.802,
      "eval_steps_per_second": 8.792,
      "step": 666
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.3895110785961151,
      "learning_rate": 1.3330645698541948e-05,
      "loss": 0.1396,
      "step": 999
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8949333333333334,
      "eval_loss": 0.35275891423225403,
      "eval_runtime": 6.8758,
      "eval_samples_per_second": 545.395,
      "eval_steps_per_second": 8.581,
      "step": 999
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.1146515607833862,
      "learning_rate": 5.006485928676918e-10,
      "loss": 0.0457,
      "step": 1332
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9048,
      "eval_loss": 0.32710474729537964,
      "eval_runtime": 7.6993,
      "eval_samples_per_second": 487.055,
      "eval_steps_per_second": 7.663,
      "step": 1332
    }
  ],
  "logging_steps": 500,
  "max_steps": 1332,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "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": 6.58965282490368e+18,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}