Search papers, labs, and topics across Lattice.
The paper introduces M3Kang, a new massively multilingual, multimodal mathematical reasoning dataset derived from the Kangaroo Math Competition, containing 1,747 problems translated into 108 languages. They benchmarked state-of-the-art VLMs on this dataset, finding that models struggle with basic math and diagram-based reasoning, and that performance scales with language presence and model size. They also demonstrate that multilingual techniques can be effectively extended to the multimodal setting, leading to performance improvements.
VLMs still can't beat middle schoolers at math, especially when diagrams and diverse languages are involved, according to a new multilingual multimodal benchmark.
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.
