trainer_state.json

{
  "best_metric": 0.9290666666666667,
  "best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/facebook_dino-vitb16/model_idx_0616/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": 0.06883314996957779,
      "learning_rate": 2.9100759384327822e-05,
      "loss": 0.7542,
      "step": 333
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8781333333333333,
      "eval_loss": 0.377939373254776,
      "eval_runtime": 10.8572,
      "eval_samples_per_second": 345.393,
      "eval_steps_per_second": 5.434,
      "step": 333
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.17441777884960175,
      "learning_rate": 2.6500767323004636e-05,
      "loss": 0.1756,
      "step": 666
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8829333333333333,
      "eval_loss": 0.36711302399635315,
      "eval_runtime": 10.7062,
      "eval_samples_per_second": 350.266,
      "eval_steps_per_second": 5.511,
      "step": 666
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.5214559435844421,
      "learning_rate": 2.25136129892785e-05,
      "loss": 0.0713,
      "step": 999
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9082666666666667,
      "eval_loss": 0.3029240667819977,
      "eval_runtime": 10.6717,
      "eval_samples_per_second": 351.396,
      "eval_steps_per_second": 5.529,
      "step": 999
    },
    {
      "epoch": 4.0,
      "grad_norm": Infinity,
      "learning_rate": 1.763568653571882e-05,
      "loss": 0.0283,
      "step": 1332
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9125333333333333,
      "eval_loss": 0.34512650966644287,
      "eval_runtime": 10.3625,
      "eval_samples_per_second": 361.883,
      "eval_steps_per_second": 5.694,
      "step": 1332
    },
    {
      "epoch": 5.0,
      "grad_norm": 1.0790977478027344,
      "learning_rate": 1.2426252654184347e-05,
      "loss": 0.0089,
      "step": 1665
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9130666666666667,
      "eval_loss": 0.35207831859588623,
      "eval_runtime": 10.4719,
      "eval_samples_per_second": 358.102,
      "eval_steps_per_second": 5.634,
      "step": 1665
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.12751924991607666,
      "learning_rate": 7.5272506870206094e-06,
      "loss": 0.0015,
      "step": 1998
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.9274666666666667,
      "eval_loss": 0.32007110118865967,
      "eval_runtime": 10.8559,
      "eval_samples_per_second": 345.435,
      "eval_steps_per_second": 5.435,
      "step": 1998
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.023093033581972122,
      "learning_rate": 3.5295725730362483e-06,
      "loss": 0.0006,
      "step": 2331
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.9277333333333333,
      "eval_loss": 0.3138616681098938,
      "eval_runtime": 10.4356,
      "eval_samples_per_second": 359.348,
      "eval_steps_per_second": 5.654,
      "step": 2331
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.046832192689180374,
      "learning_rate": 9.153972922231291e-07,
      "loss": 0.0001,
      "step": 2664
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.9290666666666667,
      "eval_loss": 0.3166520893573761,
      "eval_runtime": 10.0534,
      "eval_samples_per_second": 373.01,
      "eval_steps_per_second": 5.869,
      "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
}