izhx/UMRB-LLaVA

该数据集是一个用于信息检索和多模态检索的测试数据集，包含三个主要配置：qrels、corpus和query。qrels配置包含查询ID、Q0、语料库ID和分数等特征，用于评估检索结果的相关性。corpus配置包含ID、模态、文本和图像等特征，用于存储语料库信息。query配置包含ID、模态、文本和图像等特征，用于存储查询信息。数据集的分割信息详细列出，例如qrels的测试分割有5120个样本，corpus的语料库分割有5994个样本，query的测试分割有5120个样本。数据集使用了cc-by-4.0许可证。

The dataset consists of three main configurations: qrels, corpus, and query. The qrels configuration is used to evaluate the relationship between queries and the corpus, featuring query-id, Q0, corpus-id, and score, applicable to the test set with 5120 samples. The corpus configuration includes features such as id, modality, text, and image, totaling 5994 samples. The query configuration is similar to the corpus configuration, featuring id, modality, text, and image, applicable to the test set with 5120 samples. The dataset is primarily used for information retrieval and multimodal retrieval tasks.