alias-free-gan-pytorch
alias-free-gan-pytorch copied to clipboard
discard unusable images
discard unusable images using a try..catch and subsequent length checking
To me, it would maybe make more sense to make sure that your dataset has no broken images before running any training on it, instead of silently failing on those images. I feel like this PR adds a fix to a problem unrelated with alias-free gan, and only related to a specific dataset that should probably be fixed instead, idk 🤷
I think it will be better to specify exceptions if exception handling is needed.
apologies, I agree it is not related to alias-free-gan. I was just trying to make it easier to run in collab environments while I am doing that myself. it is an awesome repo ! kudos! feel free to reject the pull request, will specify the exceptions today.
As a software developer with 25 years of experience, here's how I see it: If we know about a potential error that may occur, we should treat it. And using try/catch is the obvious choice. So I agree with @sidphbot.