checkpoint-625/trainer_state.json

{
  "best_global_step": 625,
  "best_metric": 0.1934153288602829,
  "best_model_checkpoint": "siglip2-finetune-full/checkpoint-625",
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 625,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.8,
      "grad_norm": 16.568904876708984,
      "learning_rate": 6.249999999999999e-07,
      "loss": 0.3408,
      "step": 500
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9169958497924896,
      "eval_loss": 0.1934153288602829,
      "eval_model_preparation_time": 0.0027,
      "eval_runtime": 479.9201,
      "eval_samples_per_second": 41.672,
      "eval_steps_per_second": 5.209,
      "step": 625
    }
  ],
  "logging_steps": 500,
  "max_steps": 1250,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "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": 1.6750261609304187e+18,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}