Skip to content

edinburgh-dawg/mmlu-redux-2.0

Question AnsweringENBenchmarkcc-by-4.0

Edinburgh-dawg/mmlu-redux-2.0 is a question answering-focused benchmark dataset in EN that provides 5,700 labeled examples distributed in Parquet format. It is distributed under the cc-by-4.0 license and falls in the 1K<n<10K size category, and has been downloaded 15.2K times.

📊 This dataset is used as an LLM benchmark. See model leaderboards →

About edinburgh-dawg/mmlu-redux-2.0

Dataset Card for MMLU-Redux-2.0 MMLU-Redux is a subset of 5,700 manually re-annotated questions across 57 MMLU subjects. News [2025.02.25] We corrected one annotation in Abstract Algebra subset, as noted in the Issue #2. [2025.02...

Details

Task
Question Answering
Language
EN
Format
Parquet
Rows / instances
5700
Size
1K<n<10K
Creator
edinburgh-dawg
Year
2024
License
cc-by-4.0
Downloads
15222
Likes
37
Download Homepage

Related Question Answering datasets

FAQ