Issue with Decoding test.csv Files in PathVQA Dataset

[LiangXin1001]

Hello,

I recently downloaded the PathVQA dataset and found that test.csv, train.csv, val.csv files appear to be encoded or in a format that is not human-readable (as shown in the attached screenshots). I suspect these files might be in Base64 encoding or a similar binary-to-text encoding scheme, but I am unsure how to properly decode them for use.

Could you provide guidance on how to decode these .csv files? Are there any specific tools or scripts within the repository that I should use to process these files for further use in a project?

Thank you.