trainer_state.json

{
  "best_metric": 0.9336,
  "best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/facebook_dino-vitb16/model_idx_0380/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": 5.406577110290527,
      "learning_rate": 2.886495269365963e-05,
      "loss": 0.8407,
      "step": 333
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8813333333333333,
      "eval_loss": 0.38278132677078247,
      "eval_runtime": 13.9478,
      "eval_samples_per_second": 268.859,
      "eval_steps_per_second": 4.23,
      "step": 333
    },
    {
      "epoch": 2.0,
      "grad_norm": 15.733654022216797,
      "learning_rate": 2.5619102456015873e-05,
      "loss": 0.1968,
      "step": 666
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8893333333333333,
      "eval_loss": 0.3623836040496826,
      "eval_runtime": 12.7433,
      "eval_samples_per_second": 294.271,
      "eval_steps_per_second": 4.63,
      "step": 666
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.010589247569441795,
      "learning_rate": 2.0756590131847176e-05,
      "loss": 0.0701,
      "step": 999
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9122666666666667,
      "eval_loss": 0.3113878071308136,
      "eval_runtime": 13.7896,
      "eval_samples_per_second": 271.944,
      "eval_steps_per_second": 4.279,
      "step": 999
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.7629180550575256,
      "learning_rate": 1.5017689143724237e-05,
      "loss": 0.0281,
      "step": 1332
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9194666666666667,
      "eval_loss": 0.3023536801338196,
      "eval_runtime": 12.2889,
      "eval_samples_per_second": 305.154,
      "eval_steps_per_second": 4.801,
      "step": 1332
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.4564429223537445,
      "learning_rate": 9.276095143821772e-06,
      "loss": 0.0087,
      "step": 1665
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9250666666666667,
      "eval_loss": 0.3125024139881134,
      "eval_runtime": 12.1215,
      "eval_samples_per_second": 309.367,
      "eval_steps_per_second": 4.867,
      "step": 1665
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.022033466026186943,
      "learning_rate": 4.4059137709457125e-06,
      "loss": 0.003,
      "step": 1998
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.9293333333333333,
      "eval_loss": 0.29885414242744446,
      "eval_runtime": 12.2457,
      "eval_samples_per_second": 306.23,
      "eval_steps_per_second": 4.818,
      "step": 1998
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.013916187919676304,
      "learning_rate": 1.1485859908123557e-06,
      "loss": 0.0008,
      "step": 2331
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.9322666666666667,
      "eval_loss": 0.29362934827804565,
      "eval_runtime": 11.9913,
      "eval_samples_per_second": 312.726,
      "eval_steps_per_second": 4.92,
      "step": 2331
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.0009996391599997878,
      "learning_rate": 1.04301971493026e-11,
      "loss": 0.0006,
      "step": 2664
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.9336,
      "eval_loss": 0.2926299571990967,
      "eval_runtime": 12.5091,
      "eval_samples_per_second": 299.782,
      "eval_steps_per_second": 4.717,
      "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
}