Jian Shi

Any updates on this?

Not pretty confident. But I think this shall be the idea. ```python q_slt = torch.cat([ torch.clamp(q_slt[..., :N], 0, x.size(2) - 1), torch.clamp(q_slt[..., N:2 * N], 0, x.size(3) - 1), torch.clamp(q_slt[...,...

Your question is quite ambiguous. But I think it was probably caused by a high LR or a high loss weight which makes gradient descent optimization failed.

Generally, I think it should be alright as long as the label will not be changed by augmentation methods. For instance, ElasticTransform is probably a dangerous move. It would be...

Hi @duc12111, can you reproduce this on Colab? So that we can investigate further. It looks like an env issue.

@duc12111 You may install kornia on colab with "!pip install kornia". Then put your snippet afterwards.

@duc12111 Probably comes by the low CUDA version on your PC. It should be working. Yes. Please go for the other augmentations. Not sure what happens to the CI for...

@edgarriba It seems the os tests are not functioning? And we cannot re-run it?

@duc12111 Can you rebase to see if it will pass?

> Hi @edgarriba, I wonder if this is a mistake > > https://github.com/kornia/kornia/blob/fac687ed482fdc0493a58931e881a751a20495ad/kornia/augmentation/base.py#L219 seems to create duplicate class name since https://github.com/kornia/kornia/blob/fac687ed482fdc0493a58931e881a751a20495ad/kornia/augmentation/base.py#L62 have already had class name Yes. It is a...