Hi there, just wanted to file an issue on a small roadblock I ran into. The dataset (HEST [1]) I'm working with has some downscaled images with the extension .jpeg which trips the validity check here which only looks for .jpg. It would be convenient to add patterns for jpeg, tiff, PNG, etc. so that users don't need to rename their files. I can submit a PR if that would be helpful.
[1] https://huggingface.co/datasets/MahmoodLab/hest