mlfoundations-dev/BBEH_evalchemy

该数据集包含了一系列文本上下文信息、生成参数、索引信息、任务名称以及元数据等。具体来说，文本上下文信息包括内容和角色两种类型的数据，生成参数涉及是否抽样、生成新token的最大数量、随机种子和温度等设置。索引信息包括重复索引和请求索引。元数据则包含了期望答案、问题ID和参考解决方案等。数据集被划分为训练集，共有4520个示例，大小为19416462字节。

The dataset consists of a series of text contexts, generation parameters, index information, task names, and metadata. Specifically, the text context information includes content and role types of data, generation parameters involve whether to sample, the maximum number of new tokens to generate, random seed, and temperature settings. Index information includes repeat index and request index. Metadata contains expected answers, problem IDs, and reference solutions. The dataset is split into a training set with a total of 4,520 examples and a size of 19,416,462 bytes.