diff --git a/chinese-instruct.py b/chinese-instruct.py index 8543323..e8d551e 100644 --- a/chinese-instruct.py +++ b/chinese-instruct.py @@ -1,11 +1,20 @@ -# Mxode/Chinese-Reasoning-Distil-Data - from datasets import DatasetDict from datasets import load_dataset, concatenate_datasets from utils import process_dataset -def build_prompt(query, response) -> str: +def build_prompt(query:str, response:str) -> str: + replacements = { + "\\[": "$$", + "\\]": "$$", + "\\(": "$", + "\\)": "$" + } + + for old, new in replacements.items(): + query = query.replace(old, new) + response = response.replace(old, new) + return f"<|user|> {query} <|system|> {response}\n"