trainer_state.json

{
  "best_metric": 0.22988505747126436,
  "best_model_checkpoint": "vit-base-patch16-224-in21k-finetune/checkpoint-13",
  "epoch": 4.615384615384615,
  "eval_steps": 500,
  "global_step": 15,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.92,
      "eval_accuracy": 0.09195402298850575,
      "eval_loss": 1.6376721858978271,
      "eval_runtime": 0.7691,
      "eval_samples_per_second": 113.12,
      "eval_steps_per_second": 2.6,
      "step": 3
    },
    {
      "epoch": 1.85,
      "eval_accuracy": 0.19540229885057472,
      "eval_loss": 1.6207975149154663,
      "eval_runtime": 0.7427,
      "eval_samples_per_second": 117.142,
      "eval_steps_per_second": 2.693,
      "step": 6
    },
    {
      "epoch": 2.77,
      "eval_accuracy": 0.21839080459770116,
      "eval_loss": 1.6119964122772217,
      "eval_runtime": 0.7397,
      "eval_samples_per_second": 117.615,
      "eval_steps_per_second": 2.704,
      "step": 9
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.20820951461792,
      "learning_rate": 1.923076923076923e-05,
      "loss": 1.5539,
      "step": 10
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.22988505747126436,
      "eval_loss": 1.605418086051941,
      "eval_runtime": 0.7627,
      "eval_samples_per_second": 114.068,
      "eval_steps_per_second": 2.622,
      "step": 13
    },
    {
      "epoch": 4.62,
      "eval_accuracy": 0.22988505747126436,
      "eval_loss": 1.6043109893798828,
      "eval_runtime": 0.734,
      "eval_samples_per_second": 118.535,
      "eval_steps_per_second": 2.725,
      "step": 15
    },
    {
      "epoch": 4.62,
      "step": 15,
      "total_flos": 2.799085924820828e+17,
      "train_loss": 1.5239589055379232,
      "train_runtime": 123.7366,
      "train_samples_per_second": 31.317,
      "train_steps_per_second": 0.121
    }
  ],
  "logging_steps": 10,
  "max_steps": 15,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "total_flos": 2.799085924820828e+17,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}