hsila/chem-hotpotqa

该数据集是HotpotQA数据集的化学子集，专注于化学领域的检索任务。它包括三个主要部分：语料库（Corpus）包含464个条目，查询（Queries）共274个（分为训练集234个、开发集14个、测试集26个），以及查询相关性判断（Qrels）共548个（分为训练集468个、开发集28个、测试集52个）。数据集设计用于支持化学相关的问题回答和信息检索应用，通过结构化数据促进模型训练和评估。This dataset is a chemistry subset of HotpotQA, focused on retrieval tasks in the chemistry domain. It consists of three main components: a corpus with 464 entries, queries totaling 274 (split into 234 for training, 14 for development, and 26 for testing), and query relevance judgments (qrels) totaling 548 (split into 468 for training, 28 for development, and 52 for testing). The dataset is designed to support question answering and information retrieval applications in chemistry, facilitating model training and evaluation through structured data.