New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

F.interpolate upsample gradient with amp #75

Open

bhack opened this issue Mar 2, 2024 · 1 comment

bhack commented Mar 2, 2024

@z-x-yang What impact we have a training from pytorch/pytorch#104157?

I don't know if it is the same with bfloat16 but pytorch-correlation doesn't support bfloat:
https://github.com/ClementPinard/Pytorch-Correlation-extension/blob/master/Correlation_Module/correlation_cuda_kernel.cu#L262

Author

bhack commented Mar 4, 2024 •

edited

Loading

I have a PR to support bfloat16 in the third party pytorch correlation ops ClementPinard/Pytorch-Correlation-extension#106

But we still don't know if pytorch/pytorch#104157 it has the same problem with bfloat16.

See also pytorch/pytorch#121072

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment