trainer_state.json

{
  "best_metric": 0.9106666666666666,
  "best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/facebook_dino-vitb16/model_idx_0645/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": 96.11920166015625,
      "learning_rate": 9.623897705919314e-05,
      "loss": 0.9576,
      "step": 333
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8298666666666666,
      "eval_loss": 0.5439417958259583,
      "eval_runtime": 14.2841,
      "eval_samples_per_second": 262.53,
      "eval_steps_per_second": 4.13,
      "step": 333
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.1604273021221161,
      "learning_rate": 8.543862808774192e-05,
      "loss": 0.3633,
      "step": 666
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8261333333333334,
      "eval_loss": 0.5779120922088623,
      "eval_runtime": 12.1559,
      "eval_samples_per_second": 308.493,
      "eval_steps_per_second": 4.854,
      "step": 666
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.32690557837486267,
      "learning_rate": 6.924306924189557e-05,
      "loss": 0.2032,
      "step": 999
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8552,
      "eval_loss": 0.5504829287528992,
      "eval_runtime": 12.7904,
      "eval_samples_per_second": 293.189,
      "eval_steps_per_second": 4.613,
      "step": 999
    },
    {
      "epoch": 4.0,
      "grad_norm": 3.6298604011535645,
      "learning_rate": 5.011792754282769e-05,
      "loss": 0.116,
      "step": 1332
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8722666666666666,
      "eval_loss": 0.4727502465248108,
      "eval_runtime": 12.5565,
      "eval_samples_per_second": 298.65,
      "eval_steps_per_second": 4.699,
      "step": 1332
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.013591110706329346,
      "learning_rate": 3.0974832444380136e-05,
      "loss": 0.0547,
      "step": 1665
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8853333333333333,
      "eval_loss": 0.44551724195480347,
      "eval_runtime": 13.2007,
      "eval_samples_per_second": 284.076,
      "eval_steps_per_second": 4.469,
      "step": 1665
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.00078463030513376,
      "learning_rate": 1.472814664270234e-05,
      "loss": 0.017,
      "step": 1998
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.8989333333333334,
      "eval_loss": 0.43828627467155457,
      "eval_runtime": 13.0729,
      "eval_samples_per_second": 286.853,
      "eval_steps_per_second": 4.513,
      "step": 1998
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.005578932818025351,
      "learning_rate": 3.851280774526222e-06,
      "loss": 0.0053,
      "step": 2331
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.9056,
      "eval_loss": 0.4155564308166504,
      "eval_runtime": 13.1219,
      "eval_samples_per_second": 285.782,
      "eval_steps_per_second": 4.496,
      "step": 2331
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.016379250213503838,
      "learning_rate": 1.3906924697382196e-10,
      "loss": 0.0018,
      "step": 2664
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.9106666666666666,
      "eval_loss": 0.398611456155777,
      "eval_runtime": 13.3477,
      "eval_samples_per_second": 280.948,
      "eval_steps_per_second": 4.42,
      "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
}