mixedbread-ai/vidore-arxivqa_test_subsampled

该数据集包含三个配置：corpus、default和queries。corpus配置包含图像和corpus-id特征，default配置包含query-id、corpus-id和score特征，queries配置包含query-id和query文本特征。数据集分为训练集，每个配置的训练集包含500个样本。corpus配置的训练集大小为90193755字节，download_size为77166879字节。default配置的训练集大小为12000字节，download_size为7034字节。queries配置的训练集大小为55717字节，download_size为30828字节。

The dataset consists of three configurations: corpus, default, and queries. The corpus configuration includes image and corpus-id features, the default configuration includes query-id, corpus-id, and score features, and the queries configuration includes query-id and query text features. The dataset is split into training sets, with each configurations training set containing 500 examples. The corpus configurations training set size is 90193755 bytes, with a download size of 77166879 bytes. The default configurations training set size is 12000 bytes, with a download size of 7034 bytes. The queries configurations training set size is 55717 bytes, with a download size of 30828 bytes.