fix(chinese-instruct): 修复 LaTex 公式在数据集中的显示问题

This commit is contained in:
2025-07-21 11:36:50 +08:00
parent 127e7c8229
commit 1ef14a997e
+12 -3
View File
@@ -1,11 +1,20 @@
# Mxode/Chinese-Reasoning-Distil-Data
from datasets import DatasetDict from datasets import DatasetDict
from datasets import load_dataset, concatenate_datasets from datasets import load_dataset, concatenate_datasets
from utils import process_dataset 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|> <bos>{response}<eos>\n" return f"<|user|> {query} <|system|> <bos>{response}<eos>\n"