TAUR-dev/raw__BC__math500_gpt4omini_cot_evals

该数据集包含了文档ID、文档内容（答案、难度级别、问题、解决方案、学科、唯一ID等）、目标、参数、响应、过滤后的响应、文档哈希值、提示哈希值、目标哈希值、精确匹配、提取的答案、行为提示、行为类型、行为跨度、新行为跨度计数和行为计数等字段。数据集分为训练集，共有400个示例，文件大小为3,498,075字节。

The dataset includes fields such as document ID, document content (answer, difficulty level, problem, solution, subject, unique ID), target, parameters, response, filtered response, document hash, prompt hash, target hash, exact match, extracted answers, behavior prompt, behavior type, behavior spans, new behavior span count, and behavior count. The dataset is split into a training set with a total of 400 examples and a file size of 3,498,075 bytes.