TAUR-dev/evals__just_attempts__samples

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、科目和唯一标识符）、目标、参数（包括生成参数和采样参数）、响应、过滤后的响应、文档哈希值、提示哈希值、目标哈希值、精确匹配标志和提取的答案列表。数据集分为训练集，大小为1,807,677字节，包含500个示例。

The dataset includes document ID, document content (including answer, level, problem, solution, subject, and unique identifier), target, arguments (including generation parameters and sampling parameters), responses, filtered responses, document hash, prompt hash, target hash, exact match flag, and a list of extracted answers. The dataset is split into a training set, which is 1,807,677 bytes in size and contains 500 examples.