Nayana-cognitivelab/nayana-beir-eval-multilang_v6

该数据集包含三个部分：文档集合(corpus)、查询与相关文档的关系(qrels)、以及查询信息(queries)。文档集合中包含文档id和图像信息，用于构建文档库。查询与相关文档的关系部分提供了查询id、文档id、是否为可回答问题的标记、答案文本和相关性分数。查询信息部分则包括查询id、查询文本、查询语言、使用GPT-4O模型得到的推理过程和答案。这三个部分均提供了测试集，分别包含3000、609和238个样本。

The dataset consists of three parts: a corpus of documents, a set of query-relevant document relationships (qrels), and query information. The corpus includes document IDs and image data for building a document library. The query-relevant document relationships part provides query IDs, document IDs, a flag indicating whether the document is answerable, answer text, and relevance scores. The query information part includes query IDs, query text, query language, reasoning process obtained using the GPT-4O model, and answers. Each of these parts provides a test set containing 3000, 609, and 238 samples respectively.