trainer_state.json

{
  "best_metric": 0.9477333333333333,
  "best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/google_vit-base-patch16-224/model_idx_0744/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": 0.2045968472957611,
      "learning_rate": 6.788924172750679e-05,
      "loss": 0.7197,
      "step": 333
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9181333333333334,
      "eval_loss": 0.27964532375335693,
      "eval_runtime": 10.463,
      "eval_samples_per_second": 358.407,
      "eval_steps_per_second": 5.639,
      "step": 333
    },
    {
      "epoch": 2.0,
      "grad_norm": 3.625582456588745,
      "learning_rate": 6.181155550916422e-05,
      "loss": 0.13,
      "step": 666
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9298666666666666,
      "eval_loss": 0.2452612817287445,
      "eval_runtime": 10.2019,
      "eval_samples_per_second": 367.578,
      "eval_steps_per_second": 5.783,
      "step": 666
    },
    {
      "epoch": 3.0,
      "grad_norm": 6.099792957305908,
      "learning_rate": 5.2499999999999995e-05,
      "loss": 0.0601,
      "step": 999
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9317333333333333,
      "eval_loss": 0.2514972686767578,
      "eval_runtime": 10.6143,
      "eval_samples_per_second": 353.296,
      "eval_steps_per_second": 5.559,
      "step": 999
    },
    {
      "epoch": 4.0,
      "grad_norm": 13.129115104675293,
      "learning_rate": 4.1077686218342566e-05,
      "loss": 0.0226,
      "step": 1332
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9413333333333334,
      "eval_loss": 0.2308395504951477,
      "eval_runtime": 10.6748,
      "eval_samples_per_second": 351.296,
      "eval_steps_per_second": 5.527,
      "step": 1332
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.02299739420413971,
      "learning_rate": 2.8922313781657438e-05,
      "loss": 0.008,
      "step": 1665
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9442666666666667,
      "eval_loss": 0.22834542393684387,
      "eval_runtime": 10.5041,
      "eval_samples_per_second": 357.003,
      "eval_steps_per_second": 5.617,
      "step": 1665
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.07183569669723511,
      "learning_rate": 1.7500000000000005e-05,
      "loss": 0.003,
      "step": 1998
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.9432,
      "eval_loss": 0.22356565296649933,
      "eval_runtime": 10.3715,
      "eval_samples_per_second": 361.567,
      "eval_steps_per_second": 5.689,
      "step": 1998
    },
    {
      "epoch": 7.0,
      "grad_norm": NaN,
      "learning_rate": 8.212042196399656e-06,
      "loss": 0.0016,
      "step": 2331
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.9469333333333333,
      "eval_loss": 0.2250044047832489,
      "eval_runtime": 10.3667,
      "eval_samples_per_second": 361.734,
      "eval_steps_per_second": 5.691,
      "step": 2331
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.009271721355617046,
      "learning_rate": 2.1233245810799825e-06,
      "loss": 0.0014,
      "step": 2664
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.9477333333333333,
      "eval_loss": 0.22123831510543823,
      "eval_runtime": 11.0125,
      "eval_samples_per_second": 340.523,
      "eval_steps_per_second": 5.358,
      "step": 2664
    }
  ],
  "logging_steps": 500,
  "max_steps": 2997,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 9,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.317930564980736e+19,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}