trainer_state.json

{
  "best_metric": 0.9093333333333333,
  "best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/facebook_vit-mae-base/model_idx_0236/checkpoints/checkpoint-2997",
  "epoch": 9.0,
  "eval_steps": 500,
  "global_step": 2997,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 9.120213508605957,
      "learning_rate": 9.698463103929542e-05,
      "loss": 1.4992,
      "step": 333
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.816,
      "eval_loss": 0.6643415093421936,
      "eval_runtime": 12.3038,
      "eval_samples_per_second": 304.784,
      "eval_steps_per_second": 4.795,
      "step": 333
    },
    {
      "epoch": 2.0,
      "grad_norm": 7.678088188171387,
      "learning_rate": 8.83022221559489e-05,
      "loss": 0.5033,
      "step": 666
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8429333333333333,
      "eval_loss": 0.5406781435012817,
      "eval_runtime": 12.0047,
      "eval_samples_per_second": 312.377,
      "eval_steps_per_second": 4.915,
      "step": 666
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.7909250259399414,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.285,
      "step": 999
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8632,
      "eval_loss": 0.47522884607315063,
      "eval_runtime": 12.1298,
      "eval_samples_per_second": 309.157,
      "eval_steps_per_second": 4.864,
      "step": 999
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.08470359444618225,
      "learning_rate": 5.868240888334653e-05,
      "loss": 0.1629,
      "step": 1332
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8736,
      "eval_loss": 0.442292720079422,
      "eval_runtime": 12.667,
      "eval_samples_per_second": 296.045,
      "eval_steps_per_second": 4.658,
      "step": 1332
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.9737549424171448,
      "learning_rate": 4.131759111665349e-05,
      "loss": 0.0805,
      "step": 1665
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8858666666666667,
      "eval_loss": 0.413420706987381,
      "eval_runtime": 12.127,
      "eval_samples_per_second": 309.228,
      "eval_steps_per_second": 4.865,
      "step": 1665
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.7177073359489441,
      "learning_rate": 2.500000000000001e-05,
      "loss": 0.0318,
      "step": 1998
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.9034666666666666,
      "eval_loss": 0.392501562833786,
      "eval_runtime": 12.3989,
      "eval_samples_per_second": 302.447,
      "eval_steps_per_second": 4.759,
      "step": 1998
    },
    {
      "epoch": 7.0,
      "grad_norm": NaN,
      "learning_rate": 1.1731488851999511e-05,
      "loss": 0.0117,
      "step": 2331
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.9058666666666667,
      "eval_loss": 0.38272520899772644,
      "eval_runtime": 12.1533,
      "eval_samples_per_second": 308.559,
      "eval_steps_per_second": 4.855,
      "step": 2331
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.04322165995836258,
      "learning_rate": 3.033320830114261e-06,
      "loss": 0.0049,
      "step": 2664
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.9077333333333333,
      "eval_loss": 0.3744620382785797,
      "eval_runtime": 12.2037,
      "eval_samples_per_second": 307.283,
      "eval_steps_per_second": 4.835,
      "step": 2664
    },
    {
      "epoch": 9.0,
      "grad_norm": 0.07138283550739288,
      "learning_rate": 2.7470478758395702e-11,
      "loss": 0.0038,
      "step": 2997
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.9093333333333333,
      "eval_loss": 0.3732796907424927,
      "eval_runtime": 12.5517,
      "eval_samples_per_second": 298.764,
      "eval_steps_per_second": 4.701,
      "step": 2997
    }
  ],
  "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": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.482671885603328e+19,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}