mtabvqa/MTabVQA-Instruct

MTabVQA-Instruct是一个用于视觉问答任务的数据集，特别针对表格数据的问答。它包含多个子数据集，如MTabVQA-Atis-Instruct、MTabVQA-MiMo-Instruct、MTabVQA-Multitab-Instruct和MTabVQA-Spider-Instruct，每个子数据集都包含一个VQA.jsonl文件和相应的表格图像目录。在使用多表格指令子数据集时，需要解压table_images.zip文件以访问表格图像。

MTabVQA-Instruct is a dataset for visual question answering tasks, specifically designed for table-based questions. It includes multiple sub-datasets such as MTabVQA-Atis-Instruct, MTabVQA-MiMo-Instruct, MTabVQA-Multitab-Instruct, and MTabVQA-Spider-Instruct, each containing a VQA.jsonl file and the corresponding table image directory. When using the Multitab-Instruct sub-dataset, it is necessary to unzip the table_images.zip file to access the table images.