prepare datasets only flag

This commit is contained in:
Wing Lian
2023-04-15 16:30:55 -04:00
parent d060c803ce
commit 23938015c8

View File

@@ -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"]