trainer_state.json

{
  "best_metric": 0.9298666666666666,
  "best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/facebook_dino-vitb16/model_idx_0360/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": 1.7083463668823242,
      "learning_rate": 4.376876876876877e-05,
      "loss": 0.7497,
      "step": 333
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8821333333333333,
      "eval_loss": 0.39682498574256897,
      "eval_runtime": 5.879,
      "eval_samples_per_second": 637.864,
      "eval_steps_per_second": 10.036,
      "step": 333
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.982249736785889,
      "learning_rate": 3.7518768768768766e-05,
      "loss": 0.1901,
      "step": 666
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9082666666666667,
      "eval_loss": 0.2894798517227173,
      "eval_runtime": 6.2554,
      "eval_samples_per_second": 599.482,
      "eval_steps_per_second": 9.432,
      "step": 666
    },
    {
      "epoch": 3.0,
      "grad_norm": Infinity,
      "learning_rate": 3.128753753753754e-05,
      "loss": 0.0756,
      "step": 999
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9133333333333333,
      "eval_loss": 0.32839521765708923,
      "eval_runtime": 6.154,
      "eval_samples_per_second": 609.36,
      "eval_steps_per_second": 9.587,
      "step": 999
    },
    {
      "epoch": 4.0,
      "grad_norm": 10.71441650390625,
      "learning_rate": 2.5037537537537537e-05,
      "loss": 0.0377,
      "step": 1332
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9162666666666667,
      "eval_loss": 0.32350385189056396,
      "eval_runtime": 6.3044,
      "eval_samples_per_second": 594.824,
      "eval_steps_per_second": 9.359,
      "step": 1332
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.3541021943092346,
      "learning_rate": 1.8787537537537537e-05,
      "loss": 0.016,
      "step": 1665
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9176,
      "eval_loss": 0.33874237537384033,
      "eval_runtime": 6.3105,
      "eval_samples_per_second": 594.245,
      "eval_steps_per_second": 9.349,
      "step": 1665
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.038011353462934494,
      "learning_rate": 1.2537537537537538e-05,
      "loss": 0.0042,
      "step": 1998
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.9277333333333333,
      "eval_loss": 0.3311464786529541,
      "eval_runtime": 5.8361,
      "eval_samples_per_second": 642.551,
      "eval_steps_per_second": 10.109,
      "step": 1998
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.005365929566323757,
      "learning_rate": 6.287537537537539e-06,
      "loss": 0.0012,
      "step": 2331
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.9258666666666666,
      "eval_loss": 0.3322850167751312,
      "eval_runtime": 5.8208,
      "eval_samples_per_second": 644.24,
      "eval_steps_per_second": 10.136,
      "step": 2331
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.001798368408344686,
      "learning_rate": 3.753753753753754e-08,
      "loss": 0.0007,
      "step": 2664
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.9298666666666666,
      "eval_loss": 0.3215586841106415,
      "eval_runtime": 5.8887,
      "eval_samples_per_second": 636.816,
      "eval_steps_per_second": 10.019,
      "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
}