Trim non-pipeline input grads before caching in bwd_cache#431
Open
aditvenk wants to merge 1 commit into
Open
Conversation
The backward graph may produce gradients for inputs beyond the pipeline activations (e.g. labels when loss is fused into the last stage). get_bwd_send_ops zips bwd_cache with grad_send_info using strict=True, and grad_send_info only has entries for pipeline activation inputs, so extra grads cause a ValueError. Mirror the trimming that upstream PipelineStage does at torch/distributed/pipelining/stage.py:997. Authored with Claude. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
meta-cla
Bot
added
the
CLA Signed
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The backward graph may produce gradients for inputs beyond the pipeline activations (e.g. labels when loss is fused into the last stage). get_bwd_send_ops zips bwd_cache with grad_send_info using strict=True, and grad_send_info only has entries for pipeline activation inputs, so extra grads cause a ValueError.
Mirror the trimming that upstream PipelineStage does at torch/distributed/pipelining/stage.py:997.
Authored with Claude.