trainer_state.json

{
  "best_metric": 0.9010666666666667,
  "best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/facebook_dino-vitb16/model_idx_0527/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": 1.2428045272827148,
      "learning_rate": 6.735155628520579e-05,
      "loss": 0.8402,
      "step": 333
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8306666666666667,
      "eval_loss": 0.5576215386390686,
      "eval_runtime": 5.529,
      "eval_samples_per_second": 678.236,
      "eval_steps_per_second": 10.671,
      "step": 333
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3940715789794922,
      "learning_rate": 5.9807039661419336e-05,
      "loss": 0.2641,
      "step": 666
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8488,
      "eval_loss": 0.5173069834709167,
      "eval_runtime": 5.3673,
      "eval_samples_per_second": 698.678,
      "eval_steps_per_second": 10.993,
      "step": 666
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.1906450241804123,
      "learning_rate": 4.8470148469326895e-05,
      "loss": 0.121,
      "step": 999
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8722666666666666,
      "eval_loss": 0.4865083396434784,
      "eval_runtime": 5.809,
      "eval_samples_per_second": 645.555,
      "eval_steps_per_second": 10.157,
      "step": 999
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.0013258703984320164,
      "learning_rate": 3.508254927997938e-05,
      "loss": 0.0529,
      "step": 1332
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8706666666666667,
      "eval_loss": 0.5167363882064819,
      "eval_runtime": 5.4484,
      "eval_samples_per_second": 688.271,
      "eval_steps_per_second": 10.829,
      "step": 1332
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.001760895480401814,
      "learning_rate": 2.1682382711066092e-05,
      "loss": 0.0187,
      "step": 1665
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8914666666666666,
      "eval_loss": 0.473054438829422,
      "eval_runtime": 5.4315,
      "eval_samples_per_second": 690.412,
      "eval_steps_per_second": 10.862,
      "step": 1665
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.018276004120707512,
      "learning_rate": 1.0309702649891636e-05,
      "loss": 0.0035,
      "step": 1998
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.8984,
      "eval_loss": 0.47078263759613037,
      "eval_runtime": 5.8967,
      "eval_samples_per_second": 635.951,
      "eval_steps_per_second": 10.006,
      "step": 1998
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.01982496678829193,
      "learning_rate": 2.6958965421683553e-06,
      "loss": 0.0006,
      "step": 2331
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.9002666666666667,
      "eval_loss": 0.47548454999923706,
      "eval_runtime": 5.3943,
      "eval_samples_per_second": 695.173,
      "eval_steps_per_second": 10.937,
      "step": 2331
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.016194652765989304,
      "learning_rate": 9.734847288167535e-11,
      "loss": 0.0003,
      "step": 2664
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.9010666666666667,
      "eval_loss": 0.47520995140075684,
      "eval_runtime": 5.5182,
      "eval_samples_per_second": 679.566,
      "eval_steps_per_second": 10.692,
      "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
}