[PyTorch] Don't use autograd hook for bwd reduction#781
Merged
ksivaman merged 1 commit intoNVIDIA:mainfrom Apr 15, 2024
Merged
Conversation
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
Member
Author
|
/te-ci pytorch |
pggPL
pushed a commit
to pggPL/TransformerEngine
that referenced
this pull request
May 23, 2024
Don't use autograd hook for bwd reduction Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com> Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Introduced in #575; using
torch.autograd.graph.register_multi_grad_hookis the suspected reason for hangs in certain workloads. This PR uses a different design to achieve the reduction of amaxes for gradient tensors by directly calling it from the backward pass of the modules if needed.