checkpoint-626/trainer_state.json

{
  "best_global_step": 626,
  "best_metric": 0.7205082774162292,
  "best_model_checkpoint": "siglip2-finetune-full/checkpoint-626",
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 626,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_accuracy": 0.6746,
      "eval_loss": 0.8624296188354492,
      "eval_model_preparation_time": 0.0023,
      "eval_runtime": 129.6436,
      "eval_samples_per_second": 77.135,
      "eval_steps_per_second": 9.642,
      "step": 313
    },
    {
      "epoch": 1.5974440894568689,
      "grad_norm": 10.701898574829102,
      "learning_rate": 0.00012512479201331117,
      "loss": 1.0067,
      "step": 500
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7695,
      "eval_loss": 0.7205082774162292,
      "eval_model_preparation_time": 0.0023,
      "eval_runtime": 129.3402,
      "eval_samples_per_second": 77.315,
      "eval_steps_per_second": 9.664,
      "step": 626
    }
  ],
  "logging_steps": 500,
  "max_steps": 1252,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "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": 1.67515158859776e+18,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}