generated_data/data_transform.py

import pandas as pd

# 读取 parquet 文件
df = pd.read_parquet("/root/githubs/LightRAG/generated_data/大生_train.parquet")

# 导出为 JSON
df.to_json("/root/githubs/LightRAG/generated_data/大生_yty_train.jsonl", orient="records", force_ascii=False, lines=True)