DeepVision-103K: A Visually Diverse, Broad-Coverage, and Verifiable Mathematical Dataset for Multimodal Reasoning

2026-02-17

AI Summary

This work addresses the limitations of existing multimodal mathematical datasets within the Reinforcement Learning with Verifiable Rewards (RLVR) framework—namely their small scale, low diversity, and limited coverage—which hinder the development of large models’ visual reasoning capabilities. To overcome this, we introduce DeepVision-103K, the first large-scale, high-quality multimodal mathematics dataset that comprehensively covers K–12 mathematical concepts, integrates rich visual elements, and supports verifiable rewards. Leveraging structured mathematical content generation and multimodal alignment techniques, our dataset substantially expands the scope and utility of RLVR training data. Models trained on DeepVision-103K achieve state-of-the-art performance on multimodal mathematical benchmarks and demonstrate strong generalization and enhanced visual reflective reasoning abilities on general visual reasoning tasks.

Technology Category

• Computer Vision: Large Vision Models
• Machine Learning: Large Multimodal Models (LMMs)
• Data Mining & Knowledge Management: Mining of Visual, Multimedia & Multimodal Data

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Large pretrained models with web data
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Reinforcement Learning with Verifiable Rewards (RLVR) has been shown effective in enhancing the visual reflection and reasoning capabilities of Large Multimodal Models (LMMs). However, existing datasets are predominantly derived from either small-scale manual construction or recombination of prior resources, which limits data diversity and coverage, thereby constraining further gains in model performance. To this end, we introduce \textbf{DeepVision-103K}, a comprehensive dataset for RLVR training that covers diverse K12 mathematical topics, extensive knowledge points, and rich visual elements. Models trained on DeepVision achieve strong performance on multimodal mathematical benchmarks, and generalize effectively to general multimodal reasoning tasks. Further analysis reveals enhanced visual perception, reflection and reasoning capabilities in trained models, validating DeepVision's effectiveness for advancing multimodal reasoning. Data: \href{https://huggingface.co/datasets/skylenage/DeepVision-103K}{this url}.

Problem

• multimodal reasoning
• mathematical dataset
• visual diversity
• RLVR
• data coverage

Innovation

• DeepVision-103K
• multimodal reasoning
• Reinforcement Learning with Verifiable Rewards
• visually diverse dataset
• Large Multimodal Models