Skip to content

Make nms support tensor values for iou_threshold #9121

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.

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

Closed
wants to merge 1 commit into from

Conversation

jansel
Copy link

@jansel jansel commented Jun 26, 2025

Not sure if this is something we want to support, but it fixes pytorch/pytorch#156722

Not sure if this is something we want to support, but it fixes #156722
Copy link

pytorch-bot bot commented Jun 26, 2025

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/vision/9121

Note: Links to docs will display an error until the docs builds have been completed.

❌ 6 New Failures

As of commit 107186a with merge base 0721867 (image):

NEW FAILURES - The following jobs have failed:

This comment was automatically generated by Dr. CI and updates every 15 minutes.

@NicolasHug
Copy link
Member

NicolasHug commented Jun 26, 2025

Thanks for the PR @jansel . I'm not sure we'll want to merge this, I think users should just pass a float as you noted. I'll reply on pytorch/pytorch#156722 and ask for more details from OP, if really there is a good reason users can't call item() themselves then we can reconsider.
I'm prefer not merging this for now without learning more, because it's possible we'd have to do the same conversion for all non-tensor parameters in all ops.

@jansel
Copy link
Author

jansel commented Jun 26, 2025

Thanks, that seems fine to me. I believe this issue might be from a fuzz tester.

@jansel jansel closed this Jun 26, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

torch.compile fails with torchvision ops due to FakeTensor handling in scalar parameters
3 participants