Skip to content

Dataset · hicai-zju

SciKnowEval multi-level scientific knowledge benchmark

hicai-zju/SciKnowEval

From 10,000 to 100,000 English scientific tasks for comprehensive LLM knowledge evaluation.

What's inside

An English text benchmark designed for multi-level evaluation of scientific knowledge in large language models. It includes 10,000 to 100,000 examples covering question-answering, text classification, summarization, and feature extraction tasks. The evaluation concept is inspired by the principles of the ancient Chinese treatise Doctrine of the Mean.

What you can do with it

  • Comprehensive testing of LLM scientific knowledge and logic across different difficulty levels
  • Evaluating models on scientific text summarization and classification tasks
Tasks
Question answering, Embeddings, Text classification, Summarization
Languages
English
Size
10K–100K rows
Downloads this month
5.7K
Likes
19

How to load it

from datasets import load_dataset

ds = load_dataset("hicai-zju/SciKnowEval")