diff --git a/scripts/finetune.py b/scripts/finetune.py index 77f1b3c78..4f2a9a255 100644 --- a/scripts/finetune.py +++ b/scripts/finetune.py @@ -306,6 +306,7 @@ def setup_trainer(cfg, train_dataset, eval_dataset, model, tokenizer): def train( config: Path = Path("configs/"), + prepare_ds_only: bool = False, **kwargs, ): if Path(config).is_dir(): @@ -396,6 +397,10 @@ def train( else: dataset.save_to_disk(DEFAULT_DATASET_PREPARED_PATH) + if prepare_ds_only: + logger.info("Finished preparing dataset. Exiting...") + return + train_dataset = dataset["train"] eval_dataset = dataset["test"]