llamaindex/vdr-multilingual-test

这是一个多语言视觉文档检索基准数据集，包含15个不同的基准，用于初始评估多模态检索嵌入模型vdr-2b-multi-v1。每个语言子集都包含该语言的查询和图像，并按照页面类型分为三种不同的类别，每个类别包含1000个图像和100个查询。数据集覆盖的主题和领域广泛多样，主要目的是测试特定文档类型的检索能力。

This is a multilingual visual document retrieval benchmark dataset, containing 15 different benchmarks used to initially evaluate the multimodal retrieval embedding model vdr-2b-multi-v1. Each language subset contains queries and images in that language, divided into three different categories by the pagetype column, with each category containing 1000 images and 100 queries. The dataset covers a wide range of topics and fields, aiming to test the retrieval capabilities for specific document types.