vidore/arxivqa_test_subsampled_beir

该数据集是BEIR版本中从vidore/arxivqa_test_subsampled数据集中抽取的子集，包含三个配置：corpus、qrels和queries。corpus配置包含图像特征，并有500个测试样本。qrels配置包含查询ID、语料库ID和得分特征，同样有500个测试样本。queries配置包含查询ID和查询文本特征，也有500个测试样本。

This dataset is a subsampled version from the vidore/arxivqa_test_subsampled in the BEIR repository, containing three configurations: corpus, qrels, and queries. The corpus configuration includes an image feature with 500 test examples. The qrels configuration consists of query-id, corpus-id, and score features, also with 500 test examples. The queries configuration includes query-id and query (string) features, and 500 test examples as well.