Dataset · ceval
C-Eval Chinese examination benchmark
ceval/ceval-exam
13,948 multiple-choice questions across 52 disciplines for evaluating Chinese foundation models.
What's inside
C-Eval is a comprehensive Chinese evaluation suite for foundation models. It contains 13,948 multiple-choice questions across 52 disciplines and four difficulty levels. Each subject includes dev, val, and test splits, with the dev split providing five examples with explanations for few-shot evaluation.
What you can do with it
- Evaluate Chinese-language LLMs in few-shot mode
- Compare models across disciplines and difficulty levels
- Verify knowledge benchmarks prior to release
- Tasks
- Text classification, Multiple choice, Question answering
- Languages
- Chinese
- Size
- 10K–100K rows
- Downloads this month
- 63.5K
- Likes
- 324
How to load it
from datasets import load_dataset
ds = load_dataset("ceval/ceval-exam")