checkpoint-2000/trainer_state.json

{
  "best_metric": 0.011061140336096287,
  "best_model_checkpoint": "surface_crack_image_detection/checkpoint-2000",
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.5,
      "grad_norm": 0.03909469395875931,
      "learning_rate": 7.692307692307693e-07,
      "loss": 0.0075,
      "step": 500
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.035770125687122345,
      "learning_rate": 5.128205128205127e-07,
      "loss": 0.0059,
      "step": 1000
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.998,
      "eval_loss": 0.013308565132319927,
      "eval_runtime": 90.6031,
      "eval_samples_per_second": 88.297,
      "eval_steps_per_second": 11.037,
      "step": 1000
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.03578406199812889,
      "learning_rate": 2.5641025641025636e-07,
      "loss": 0.0076,
      "step": 1500
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.3025926351547241,
      "learning_rate": 0.0,
      "loss": 0.0053,
      "step": 2000
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.999125,
      "eval_loss": 0.011061140336096287,
      "eval_runtime": 78.7393,
      "eval_samples_per_second": 101.601,
      "eval_steps_per_second": 12.7,
      "step": 2000
    }
  ],
  "logging_steps": 500,
  "max_steps": 2000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 500,
  "total_flos": 4.959487335333888e+18,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}