The dataset contains a single xlsx file, which is the dataset. The first row is the original problem, and the second row is its answer; the third row is the corresponding modified problem, and the fourth row is the answer. The fifth row is tag to the questions.

All images for image-based questions are inserted in the xlsx file. For image-based questions, each question takes up two rows, where the first row is the original and modified image and the second row is the text question and answers.