Skip to content

Dataset · HAERAE-HUB

Korean KMMLU benchmark

HAERAE-HUB/KMMLU

35,030 expert multiple-choice questions across 45 subjects in Korean.

What's inside

KMMLU is a Korean benchmark consisting of 35,030 multiple-choice questions across 45 subjects, ranging from humanities to STEM. The questions are gathered from original Korean exams rather than translated from English benchmarks, thereby reflecting the language and culture. The authors tested 26 public and proprietary LLMs.

What you can do with it

  • Evaluating LLMs in Korean
  • Comparing models on subject-matter knowledge
  • Checking cultural and linguistic adaptation
Tasks
Multiple choice
Languages
Korean
Size
100K–1M rows
Downloads this month
13.4K
Likes
106

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("HAERAE-HUB/KMMLU")