mteb/VidoreInfoVQARetrieval
收藏Hugging Face2025-10-21 更新2025-10-25 收录
下载链接:
https://hf-mirror.com/datasets/mteb/VidoreInfoVQARetrieval
下载链接
链接失效反馈官方服务:
资源简介:
VidoreInfoVQARetrieval是一个视觉文档检索任务的数据集,属于MTEB(Massive Text Embedding Benchmark)的一部分。数据集包含图像和文本信息,适用于学术领域。它来源于vidore/infovqa_test_subsampled_beir数据集,并在MTEB框架下进行了额外的处理。数据集分为测试集,包含994个样本,其中500个相关文档,适用于评估文本嵌入模型的性能。
VidoreInfoVQARetrieval is a visual document retrieval task dataset, part of the MTEB (Massive Text Embedding Benchmark). The dataset contains image and text information, suitable for academic fields. It is sourced from the vidore/infovqa_test_subsampled_beir dataset and has undergone additional processing within the MTEB framework. The dataset is split into a test set with 994 samples, including 500 relevant documents, for evaluating the performance of text embedding models.
提供机构:
mteb



