Nayana-cognitivelab/nayana-beir-eval-multilang_v4

该数据集包含三个部分：文档集合(corpus)、查询与文档相关性(qrels)以及查询信息(queries)。文档集合部分包含文档的标识、图片和文档ID；查询与文档相关性部分包含查询ID、文档ID、是否可回答的标记、答案和相关分数；查询信息部分包含查询ID、查询内容、查询使用的语言、GPT-4的推理结果和答案。数据集分为测试集(test)。

The dataset consists of three parts: a corpus of documents, query-document relevance, and query information. The corpus part includes document identifiers, images, and document IDs; the query-document relevance part includes query IDs, document IDs, answerability labels, answers, and relevance scores; the query information part includes query IDs, query text, the language used for the query, GPT-4 reasoning results, and answers. The dataset is split into a test set.