DocHaystack-1000

该数据集为评估大型多模态模型在涉及大量图像的视觉文档检索和理解任务上提供了一个基准。评估指标包括召回率@1、召回率@3和召回率@5。该数据集规模宏大，包含1000多个文档，专注于视觉文档检索与理解任务。This dataset provides a benchmark for evaluating large multimodal models on visual document retrieval and understanding tasks involving a large number of images. The evaluation metrics include Recall@1, Recall@3, and Recall@5. This dataset is large-scale, contains over 1,000 documents, and focuses on visual document retrieval and understanding tasks.