Jamil Zakirov

No, no quick examples, I'm using it in a different domain, not NLP tasks. I'll add tests a bit later, but for now you can adjust your example on README...

Looks like `delta` value mostly depends on top achievable performance. For example if your regular AUC is bouncing near ~0.75 it doesn't make sense to have big value, while if...

@brandenkmurray Yes, your fix is valid. I also did something similar in code I use. That's often happens at the very begging when AUC is ~0.5-0.6. Also reducing `delta` value...

I have one in private repo 😞 Will share at mid-August, as far as competition I'm taking part in will end. Nothing really serious here. Just an indexing error I...

Hi I actually never used this class for training, ended up using another approach, so the code is definitely needs some debugging. Try using hint from pytorch `torch.autograd.set_detect_anomaly(True)` I can...

@Dipet I'd like to work on this one. Any ideas how to better implement switching between different samplings? Ideally, this option should be available for all transforms out of the...

Found repo, which basically solves the same problem - [correct White-Balance augmentation](https://github.com/mahmoudnafifi/WB_color_augmenter). Maybe will just port Python code from there (MIT license allows this) 🤔

What library version did you use? Post output of `import albumentations; print(albumentations.__version__)` Tried to replicate this issue on 1.1.0 and everything worked fine (except that `A.IAAAffine` is marked as deprecated...

There is a separate GitHub repo dedicated to this problem (and working solution!): https://github.com/assafshocher/ResizeRight

Hi @leihuayi Same question here. If you didn't start working on MAD I'll assign it to another student 😄