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.

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

[FEATURE]: support multiple (partial) backward passes for zero #5601

Closed
ver217 opened this issue Apr 16, 2024 · 0 comments · Fixed by #5596
Closed

[FEATURE]: support multiple (partial) backward passes for zero #5601

ver217 opened this issue Apr 16, 2024 · 0 comments · Fixed by #5596
Assignees
Labels
enhancement New feature or request

Comments

@ver217
Copy link
Member

ver217 commented Apr 16, 2024

Describe the feature

In some vae training, users may use weight adaptive loss which may compute grad of some parameters twice, like
image

This will trigger backward hook twice.

Based on pytorch's document, we may use post-grad-accumulation hook to solve this problem.

image

@ver217 ver217 added the enhancement New feature or request label Apr 16, 2024
@ver217 ver217 self-assigned this Apr 16, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
enhancement New feature or request
Projects
None yet
Development

Successfully merging a pull request may close this issue.

1 participant