happy8825/MMLongBench_var1_2turn_fixed_retrieval
收藏Hugging Face2025-12-19 更新2025-12-20 收录
下载链接：
https://hf-mirror.com/datasets/happy8825/MMLongBench_var1_2turn_fixed_retrieval
下载链接
链接失效反馈官方服务：
资源简介：
MMLongBench是一个多模态长文档基准测试数据集，包含1073个样本，用于评估模型在处理不同类型证据（纯文本、图表、表格等）和多页文档时的性能。数据集包含多种特征，如相关页面、证据页面、问题、答案、文档类型等，旨在测试模型在复杂多模态信息下的理解和回答能力。
MMLongBench is a multimodal long-document benchmark dataset comprising 1073 samples designed to evaluate model performance across various evidence types (plain-text, figures, tables, etc.) and multi-page documents. The dataset includes features such as relevant pages, evidence pages, questions, answers, and document types, aiming to test models comprehension and response capabilities in complex multimodal contexts.
提供机构：
happy8825
