copenlu/druid

DRUID数据集包含现实世界中的（查询，上下文）对，用于研究在真实世界 Retrieval-Augmented Generation (RAG) 场景中上下文的使用和失败情况。该数据集基于自动断言验证的典型任务，对于自动化检索现实世界证据至关重要。数据集包含两个版本：DRUID和DRUID+。DRUID是DRUID+的一个高质量子集，已经手动标注了证据的相关性和立场。

The DRUID dataset contains real-world (query, context) pairs to facilitate studies of context usage and failures in real-world Retrieval-Augmented Generation (RAG) scenarios. The dataset is based on the prototypical task of automated claim verification, for which automated retrieval of real-world evidence is crucial. The dataset comes in two versions: DRUID and DRUID+. DRUID is a high-quality subset of DRUID+ that has been manually annotated for evidence relevance and stance.