haowu89/math-ai-bench-sources-high-wo-answer

该数据集包含多个配置，每个配置针对不同的数学和科学问题集（如AIME、APEX、arXiv数学、CMIMC、GPQA钻石、HMMT、IMO基准、奥林匹克竞赛基准和定理QA）。每个配置包含问题、原始解决方案、答案、来源、索引、模型、生成的解决方案列表、计数和上下文元数据（如正确性、令牌数量、正确和错误数量等）。数据集用于训练和评估模型在解决数学和科学问题上的能力。

This dataset includes multiple configurations, each targeting different sets of mathematical and scientific problems (e.g., AIME, APEX, arXiv math, CMIMC, GPQA diamond, HMMT, IMO benchmark, Olympiad benchmark, and Theorem QA). Each configuration contains problems, original solutions, answers, sources, indexes, models, lists of generated solutions, counts, and context metadata (such as correctness, token counts, numbers of correct and wrong answers, etc.). The dataset is used to train and evaluate models capabilities in solving mathematical and scientific problems.