Dataset · macabdul9
Complex text-only academic benchmark
macabdul9/hle_text_only
Text version of Humanity's Last Exam with 3,000 expert-level questions across sciences to evaluate advanced LLMs.
What's inside
It gathers 3,000 closed questions from dozens of fields, including mathematics, natural sciences, and humanities. The data was compiled by subject matter experts worldwide with contributions from the Center for AI Safety and Scale AI. Metadata lists text and image modalities, but this build is a text variant (1k to 10k entries).
What you can do with it
- Evaluating frontier knowledge and reasoning abilities in strong LLMs
- Comparing language models on complex multidisciplinary tasks without visual input
- Size
- 1K–10K rows
- Downloads this month
- 35K
- Likes
- 4
How to load it
from datasets import load_dataset
ds = load_dataset("macabdul9/hle_text_only")