M3Kang: Evaluating Multilingual Multimodal Mathematical Reasoning in Vision-Language Models

TL;DR

This paper introduces M3Kang, a large multilingual multimodal dataset for mathematical reasoning in vision-language models, revealing current models' limitations and the benefits of multilingual techniques, with extensive benchmarking and human comparison.

Contribution

The paper presents M3Kang, the first large-scale multilingual multimodal mathematical reasoning dataset, and demonstrates its utility in benchmarking and improving vision-language models.

Findings

Models struggle with basic math and diagram reasoning.

Multilingual techniques improve model performance.

Model performance correlates with language presence and size.

Abstract

Despite state-of-the-art vision-language models (VLMs) have demonstrated strong reasoning capabilities, their performance in multilingual mathematical reasoning remains underexplored, particularly when compared to human performance. To bridge this gap, we introduce M3Kang, the first massively multilingual, multimodal mathematical reasoning dataset for VLMs. It is derived from the Kangaroo Math Competition, the world's largest mathematics contest, which annually engages over six million participants under the age of 18 across more than 90 countries. M3Kang includes 1,747 unique multiple-choice problems organized by grade-level difficulty, with translations into 108 culturally diverse languages, some of them including diagrams essential for solving them. Using this dataset, we conduct extensive benchmarking on both closed- and open-source SOTA models. We observe that, despite recent advances, models still struggle with basic math and diagram-based reasoning, with performance scaling with language presence and model size, but not with grade level. We also find that multilingual techniques can be effectively extended to the multimodal setting, resulting in significant improvements over baseline approaches. Our analysis also incorporates performance data from over 68,000 students, enabling direct comparison with human performance. We are open-sourcing M3Kang, including the English-only subset M2Kang, along with the framework and codebase used to construct the dataset.