laubonghaudoi/

shunpao

本數據集收錄《申報》上海版、漢口版及香港版文章，共 2,915,397 篇。數據全部來源於 https://www.dhcdb.com.tw/SP/ 。Parquet 數據表中每行為一篇文章；原始逐日 Markdown 亦完整保留，方便閱讀、校核及重新產生數據集。

from datasets import load_dataset

dataset = load_dataset("laubonghaudoi/shunpao", split="train") print(dataset[0])

如想避免下載成個數據集，可用 streaming： from datasets import load_dataset

dataset = load_dataset( "laubonghaudoi/shunpao", split="train", streaming=True, ) first_article = next(iter(dataset))… See the full description on the dataset page: https://huggingface.co/datasets/laubonghaudoi/shunpao.