dadakid/msmarco

BEIR是一个异构基准，由代表9个信息检索任务的18个不同数据集构建而成。该`msmarco`子集是BEIR的一部分。所有任务均为英语（`en`）。数据集使用标准的BEIR检索布局，包括：`corpus`（每行一个文档，包含`_id`、`title`、`text`）和`queries`（每行一个查询，包含`_id`、`title`、`text`）。数据字段包括`_id`（唯一标识符）、`title`（标题，不可用时为空字符串）和`text`（文档/查询文本）。BEIR is a heterogeneous benchmark built from 18 diverse datasets representing 9 information retrieval tasks. This `msmarco` subset is part of BEIR. All tasks are in English (`en`). The dataset uses the standard BEIR retrieval layout and includes: `corpus` (one row per document with `_id`, `title`, `text`) and `queries` (one row per query with `_id`, `title`, `text`). Data fields include `_id` (unique identifier), `title` (title, empty string when unavailable), and `text` (document/query text).