trainer_state.json

{
  "best_metric": 0.4594666666666667,
  "best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/facebook_dino-vitb16/model_idx_0552/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": Infinity,
      "learning_rate": 0.0002887841425059038,
      "loss": 3.3343,
      "step": 333
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.18453333333333333,
      "eval_loss": 3.001511573791504,
      "eval_runtime": 13.7488,
      "eval_samples_per_second": 272.751,
      "eval_steps_per_second": 4.291,
      "step": 333
    },
    {
      "epoch": 2.0,
      "grad_norm": 18.44840431213379,
      "learning_rate": 0.0002564405961135417,
      "loss": 2.7662,
      "step": 666
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.27626666666666666,
      "eval_loss": 2.708169460296631,
      "eval_runtime": 13.4244,
      "eval_samples_per_second": 279.342,
      "eval_steps_per_second": 4.395,
      "step": 666
    },
    {
      "epoch": 3.0,
      "grad_norm": 23.32195281982422,
      "learning_rate": 0.0002078924338492993,
      "loss": 2.5168,
      "step": 999
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.32266666666666666,
      "eval_loss": 2.431225061416626,
      "eval_runtime": 13.7291,
      "eval_samples_per_second": 273.142,
      "eval_steps_per_second": 4.297,
      "step": 999
    },
    {
      "epoch": 4.0,
      "grad_norm": 8.440217018127441,
      "learning_rate": 0.00015053067332772077,
      "loss": 2.2484,
      "step": 1332
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.3661333333333333,
      "eval_loss": 2.2900161743164062,
      "eval_runtime": 13.5937,
      "eval_samples_per_second": 275.864,
      "eval_steps_per_second": 4.34,
      "step": 1332
    },
    {
      "epoch": 5.0,
      "grad_norm": 29.284475326538086,
      "learning_rate": 9.308812260256249e-05,
      "loss": 2.0043,
      "step": 1665
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.40826666666666667,
      "eval_loss": 2.1208577156066895,
      "eval_runtime": 13.3139,
      "eval_samples_per_second": 281.66,
      "eval_steps_per_second": 4.431,
      "step": 1665
    },
    {
      "epoch": 6.0,
      "grad_norm": 12.010164260864258,
      "learning_rate": 4.4309889303710674e-05,
      "loss": 1.7588,
      "step": 1998
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.4381333333333333,
      "eval_loss": 1.9965362548828125,
      "eval_runtime": 13.1322,
      "eval_samples_per_second": 285.558,
      "eval_steps_per_second": 4.493,
      "step": 1998
    },
    {
      "epoch": 7.0,
      "grad_norm": 1.8253724575042725,
      "learning_rate": 1.1622017275129708e-05,
      "loss": 1.5311,
      "step": 2331
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.4570666666666667,
      "eval_loss": 1.9900010824203491,
      "eval_runtime": 13.1302,
      "eval_samples_per_second": 285.601,
      "eval_steps_per_second": 4.493,
      "step": 2331
    },
    {
      "epoch": 8.0,
      "grad_norm": 53.01565170288086,
      "learning_rate": 9.387168731500582e-10,
      "loss": 1.3802,
      "step": 2664
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.4594666666666667,
      "eval_loss": 2.0325770378112793,
      "eval_runtime": 13.3029,
      "eval_samples_per_second": 281.894,
      "eval_steps_per_second": 4.435,
      "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
}