knowurknottty/MathNet

MathNet v0是一个高质量、大规模、多模态、多语言的数据集，专注于奥林匹克数学推理与检索。该数据集包含47个国家、17种语言的30,676个专家编写的数学问题及其解决方案，涵盖1985-2025年的国际和国家竞赛材料。数据集支持三种基准任务：问题解决、数学感知检索和检索增强的问题解决。数据集通过多阶段LLM管道进行提取和验证，确保数据质量。此外，数据集还包含详细的主题分类和丰富的元数据，适用于奥林匹克级推理评估、多语言数学评估等用途。

MathNet v0 is a high-quality, large-scale, multimodal, and multilingual dataset focused on Olympiad math reasoning and retrieval. It comprises 30,676 expert-authored problems with solutions across 47 countries and 17 languages, spanning two decades of international and national competitions. The dataset supports three benchmark tasks: problem solving, math-aware retrieval, and retrieval-augmented problem solving. It is built through a multi-stage LLM pipeline ensuring data quality and includes detailed topic taxonomy and rich metadata. The dataset is suitable for Olympiad-level reasoning evaluation, multilingual math assessment, and other applications.