diff --git a/src/ctx_to_lora/data/processing.py b/src/ctx_to_lora/data/processing.py index add15bc..8ce1424 100644 --- a/src/ctx_to_lora/data/processing.py +++ b/src/ctx_to_lora/data/processing.py @@ -37,7 +37,6 @@ def get_repeat_prompt(): return random.choice(REPEAT_PROMPTS) -# TODO: dont use closed_qa_prompting when training on self-generated responses def get_preprocessing_fn( ds_name: str, is_eval: bool ) -> Callable[[dict[str, Any]], dict[str, Any]]: @@ -77,7 +76,7 @@ def get_preprocessing_fn( "response": sample["answer"], } - elif "squad" in ds_name: + elif ds_name.startswith("squad"): def f(sample): q = sample["question"]