izhx/UMRB-OKVQA

该数据集涉及信息检索和多模态检索，包含三个主要配置：query、corpus和qrels。query配置包含查询的ID、模态、文本和图像特征，测试集有5046个样本。corpus配置包含语料库的ID、模态、文本和图像特征，语料库集有114516个样本。qrels配置包含查询ID、Q0、语料库ID和分数特征，测试集有37604个样本。数据集可能与视觉问答任务相关，特别是OK-VQA基准。

This dataset involves information retrieval and multimodal retrieval, consisting of three main configurations: query, corpus, and qrels. The query configuration includes features such as query ID, modality, text, and image, with a test set of 5046 examples. The corpus configuration includes features such as corpus ID, modality, text, and image, with a corpus set of 114516 examples. The qrels configuration includes features such as query ID, Q0, corpus ID, and score, with a test set of 37604 examples. The dataset is likely related to visual question answering tasks, particularly the OK-VQA benchmark.