Skip to content

Dataset · OpenStellarTeam

Chinese factual accuracy benchmark for LLMs

OpenStellarTeam/Chinese-SimpleQA

First comprehensive Chinese-language benchmark for evaluating model factual knowledge using short-form questions across 99 subtopics.

What's inside

Contains short Chinese questions and answers to test the factual accuracy of language models (question-answering). Dataset size is 1 to 10 thousand examples. Questions cover 6 major topics and 99 subtopics, selected so answers remain static and easily verifiable.

What you can do with it

  • Evaluating factual accuracy and hallucination levels of LLMs in Chinese
  • Comparing models on short-form encyclopedic question answering
Tasks
Question answering
Languages
Chinese
Size
1K–10K rows
Downloads this month
3.8K
Likes
38

How to load it

from datasets import load_dataset

ds = load_dataset("OpenStellarTeam/Chinese-SimpleQA")