This dataset is a multilingual version of the original MMLU (Measuring Massive Multitask Language Understanding) dataset, which consists of multiple-choice questions designed to test the reasoning abilities of AI systems. The polyglot version includes translations of the original English questions into various languages, allowing for evaluation of language models across different linguistic contexts.
All languages supported
- ar (Arabic)
- bn (Bengali)
- ca (Catalan)… See the full description on the dataset page: https://huggingface.co/datasets/Polygl0t/MMLU-poly.
