TAUR-dev/raw__BC__math500_r1_evals

该数据集包含文档ID、文档内容（包括答案、难度等级、问题、解决方案、主题、唯一标识符等字段）、目标标签、参数（包括生成参数等）、响应和过滤后的响应、文档哈希、提示哈希、目标哈希、精确匹配标识、提取的答案、行为提示、行为类型、行为跨度信息、新的行为跨度计数和行为计数等。数据集分为训练集，大小为16948858字节，共有400个示例。

The dataset includes document ID, document content (including fields such as answer, difficulty level, problem, solution, subject, unique identifier), target label, arguments (including generation parameters), responses and filtered responses, document hash, prompt hash, target hash, exact match indicator, extracted answers, behavior prompt, behavior type, behavior span information, new behavior span count, and behavior count. The dataset is split into a training set, which is 16948858 bytes in size and contains 400 examples.