checkpoint-1378/trainer_state.json

{
  "best_global_step": 1378,
  "best_metric": 0.34233906865119934,
  "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": 9.84671688079834,
      "learning_rate": 0.00016674057649667407,
      "loss": 0.9043,
      "step": 500
    },
    {
      "epoch": 0.7256894049346879,
      "grad_norm": 11.858268737792969,
      "learning_rate": 0.00012978566149297858,
      "loss": 0.5152,
      "step": 1000
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9016836086404066,
      "eval_loss": 0.34233906865119934,
      "eval_model_preparation_time": 0.0023,
      "eval_runtime": 552.3647,
      "eval_samples_per_second": 79.788,
      "eval_steps_per_second": 9.973,
      "step": 1378
    }
  ],
  "logging_steps": 500,
  "max_steps": 2756,
  "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": 3.692588424596177e+18,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}