mixedbread-ai/vidore-docvqa_test_subsampled

该数据集包含三个部分：corpus、default和queries。corpus部分包含图像和对应的标识id，default部分包含查询id、语料库id和评分，queries部分包含查询id和查询文本。数据集分为训练集，每个部分的大小和示例数量都有所不同。

The dataset consists of three parts: corpus, default, and queries. The corpus part includes images and their corresponding id, the default part includes query id, corpus id, and score, and the queries part includes query id and query text. The dataset is split into training sets, with each part having different sizes and numbers of examples.