michaeldinzinger/msmarco-document

该数据集是一个单语言（英语）的文本检索数据集，源数据集为msmarco。数据集包含三个配置：default、corpus和queries。default配置用于开发集，包含query-id、corpus-id和score特征，共有5193个示例。corpus配置包含_id、title和text特征，共有3213835个示例。queries配置包含_id和text特征，共有5193个示例。

This dataset is used for text retrieval tasks and includes three configurations: default, corpus, and queries. The default configuration is used to evaluate the relevance between queries and documents, containing features such as query-id, corpus-id, and score. The corpus configuration is used to store document content, including document _id, title, and text. The queries configuration is used to store query content, including query _id and text. The dataset originates from MSMARCO.