ManyModalQA

该数据集使用了代码片段、图片和表格，但更加强调在回答问题时对模态选择的重要性，而非知识整合。此外，该数据集还专注于回答问题时的模态选择，其任务是视觉问答。

This dataset incorporates code snippets, images and tables, but places greater emphasis on the significance of modality selection during question answering rather than knowledge integration. Furthermore, this dataset focuses specifically on modality selection in question answering, with its task being visual question answering (VQA).