checkpoint-582/trainer_state.json

{
  "best_global_step": 582,
  "best_metric": 0.2726958692073822,
  "best_model_checkpoint": "siglip2-finetune-full/checkpoint-582",
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 582,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7899548289954829,
      "eval_loss": 0.7594003081321716,
      "eval_model_preparation_time": 0.0023,
      "eval_runtime": 118.1364,
      "eval_samples_per_second": 78.706,
      "eval_steps_per_second": 9.845,
      "step": 291
    },
    {
      "epoch": 1.718213058419244,
      "grad_norm": 9.803659439086914,
      "learning_rate": 9.064398541919807e-05,
      "loss": 1.3365,
      "step": 500
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9234243923424392,
      "eval_loss": 0.2726958692073822,
      "eval_model_preparation_time": 0.0023,
      "eval_runtime": 117.6867,
      "eval_samples_per_second": 79.006,
      "eval_steps_per_second": 9.882,
      "step": 582
    }
  ],
  "logging_steps": 500,
  "max_steps": 873,
  "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": 1.558266304804651e+18,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}