MathCoder-VL: Bridging Vision and Code for Enhanced Multimodal Mathematical Reasoning
MathCoder-VL: Bridging Vision and Code for Enhanced Multimodal Mathematical Reasoning
Natural language image-caption datasets, widely used for training Large Multimodal Models, mainly focus on natural scenarios and overlook the intricate details of mathematical figures that are critical for problem-solving, hindering the advancement of current LMMs in multimodal mathematical reasoning. To this end, we propose leveraging code as supervision for cross-modal alignment, since code inherently encodes all information needed to generate corresponding figures, establishing a precise connection between the two modalities. Specifically, we co-develop our image-to-code model and dataset with model-in-the-loop approach, resulting in an image-to-code model, FigCodifier and ImgCode-8.6M dataset, the largest image-code dataset to date. Furthermore, we utilize FigCodifier to synthesize novel mathematical figures and then construct MM-MathInstruct-3M, a high-quality multimodal math instruction fine-tuning dataset. Finally, we present MathCoder-VL, trained with ImgCode-8.6M for cross-modal alignment and subsequently fine-tuned on MM-MathInstruct-3M for multimodal math problem solving. Our model achieves a new open-source SOTA across all six metrics. Notably, it surpasses GPT-4o and Claude 3.5 Sonnet in the geometry problem-solving subset of MathVista, achieving improvements of 8.9% and 9.2%. The dataset and models will be released at https://github.com/mathllm/MathCoder.
Houxing Ren、Mingjie Zhan、Hongsheng Li、Ke Wang、Junting Pan、Linda Wei、Aojun Zhou、Weikang Shi、Zimu Lu、Han Xiao、Yunqiao Yang
计算技术、计算机技术
Houxing Ren,Mingjie Zhan,Hongsheng Li,Ke Wang,Junting Pan,Linda Wei,Aojun Zhou,Weikang Shi,Zimu Lu,Han Xiao,Yunqiao Yang.MathCoder-VL: Bridging Vision and Code for Enhanced Multimodal Mathematical Reasoning[EB/OL].(2025-05-15)[2025-07-01].https://arxiv.org/abs/2505.10557.点此复制
评论