0

MultiBLiMP 1.0: A Massively Multilingual Benchmark of Linguistic Minimal Pairs

MultiBLiMP 1.0 evaluates large language models across 101 languages and 6 linguistic phenomena using a large dataset of minimal pairs, revealing challenges in modeling low-resource languages.

Year
2025
Venue
arXiv 2025
Authors
3
Hosting
Abstract onlyARXIV-DEFAULT

Cite

Notes

Only stored in your browser.

Attribution

Abstract & full text
arxiv.org/abs/2504.02768ARXIV-DEFAULT
TL;DR
Semantic Scholar
Attribution policy →

Abstract

We introduce MultiBLiMP 1.0, a massively multilingual benchmark of linguistic minimal pairs, covering 101 languages, 6 linguistic phenomena and containing more than 125,000 minimal pairs. Our minimal pairs are created using a fully automated pipeline, leveraging the large-scale linguistic resources of Universal Dependencies and UniMorph. MultiBLiMP 1.0 evaluates abilities of LLMs at an unprecedented multilingual scale, and highlights the shortcomings of the current state-of-the-art in modelling low-resource languages.

Authors

3