checkpoint-1378/trainer_state.json

{
  "best_global_step": 1378,
  "best_metric": 0.7454163432121277,
  "best_model_checkpoint": "siglip2-finetune-full/checkpoint-1378",
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 1378,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.36284470246734396,
      "grad_norm": 22.187894821166992,
      "learning_rate": 8.898139079333986e-07,
      "loss": 0.9313,
      "step": 500
    },
    {
      "epoch": 0.7256894049346879,
      "grad_norm": 24.758203506469727,
      "learning_rate": 7.67384916748286e-07,
      "loss": 0.7624,
      "step": 1000
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.6740333998910873,
      "eval_loss": 0.7454163432121277,
      "eval_model_preparation_time": 0.0023,
      "eval_runtime": 535.389,
      "eval_samples_per_second": 82.318,
      "eval_steps_per_second": 10.29,
      "step": 1378
    }
  ],
  "logging_steps": 500,
  "max_steps": 4134,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.69133343103497e+18,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}