Dataset · hfl
Chinese Wikipedia Question Answering
hfl/cmrc2018
About 20,000 question-answer pairs with span-based answers for Chinese reading comprehension.
What's inside
It contains nearly 20,000 real Chinese questions created by experts based on Wikipedia paragraphs, where the answer is an exact span from the source text. It also includes a challenging set with questions requiring deep understanding and logical reasoning across multiple sentences.
What you can do with it
- Training and evaluating Chinese extractive question-answering systems
- Testing LLMs' ability to perform close reading and multi-hop reasoning over text
- Tasks
- Question answering
- Languages
- Chinese
- Size
- 10K–100K rows
- Downloads this month
- 2.9K
- Likes
- 39
Models trained on it
How to load it
from datasets import load_dataset
ds = load_dataset("hfl/cmrc2018")