Nayana-cognitivelab/nayana-beir-eval-multilang_v2

该数据集包含三个部分：文档集合(corpus)、查询与答案关系(qrels)、查询信息(queries)。文档集合部分包含文档id和对应的图像；查询与答案关系部分包含查询id、文档id、是否为答案的标记、答案文本以及相关分数；查询信息部分包含查询id、查询文本、查询语言、GPT-4O的推理过程以及答案。测试集包含3000个文档集合示例、1800个查询与答案关系示例和600个查询信息示例。

The dataset consists of three parts: document collection (corpus), query-answer relationship (qrels), and query information (queries). The document collection part includes document ids and corresponding images; the query-answer relationship part includes query ids, document ids, answerability labels, answer text, and related scores; the query information part includes query ids, query text, query language, GPT-4O reasoning process, and answers. The test set contains 3000 examples of document collections, 1800 examples of query-answer relationships, and 600 examples of query information.