Skip to content
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

weights_only default flip for torch.load #1356

Closed
mikaylagawarecki opened this issue Nov 7, 2024 · 1 comment
Closed

weights_only default flip for torch.load #1356

mikaylagawarecki opened this issue Nov 7, 2024 · 1 comment
Assignees
@Jack-Khuu
Labels
actionable Items in the backlog waiting for an appropriate impl/fix enhancement New feature or request triaged This issue has been looked at a team member, and triaged and prioritized into an appropriate module

Comments

@mikaylagawarecki
Copy link

mikaylagawarecki commented Nov 7, 2024 •
edited
Loading

🐛 Describe the bug

We've flipped the default for the weights_only argument in torch.load to True in pytorch/pytorch, see here for details + documentation and this is coming in torch 2.6.

This is expected to be quite a BC-breaking change, especially if any torch.load calls in torchchat are not loading state_dicts of plain tensors.

We should make sure that all the torch.load calls in torchchat are still working.

For example, this one does not explicitly set weights_only so might now be failing since torchchat doesn't use pytorch nightlies in CI

cc @Jack-Khuu

Versions

main
@mikaylagawarecki mikaylagawarecki changed the title weights_only default flip weights_only default flip for torch.load Nov 8, 2024
@Jack-Khuu Jack-Khuu added enhancement New feature or request actionable Items in the backlog waiting for an appropriate impl/fix labels Nov 9, 2024
This was referenced Nov 12, 2024
Closed
Merged
@Jack-Khuu Jack-Khuu self-assigned this Nov 14, 2024
@Jack-Khuu Jack-Khuu added the triaged This issue has been looked at a team member, and triaged and prioritized into an appropriate module label Dec 17, 2024
@Jack-Khuu
Copy link
Contributor

Fixed in pin bump

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

@Jack-Khuu Jack-Khuu

Labels
actionable Items in the backlog waiting for an appropriate impl/fix enhancement New feature or request triaged This issue has been looked at a team member, and triaged and prioritized into an appropriate module
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@Jack-Khuu @mikaylagawarecki