Empty grad fix #291
Merged
Empty grad fix #291
Conversation
This file contains 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
frankseide
reviewed
Jul 15, 2020
samyam
reviewed
Jul 15, 2020
samyam
approved these changes
Jul 15, 2020
tjruwase
reviewed
Jul 15, 2020
tjruwase
approved these changes
Jul 15, 2020
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
This fixes a case where there is an imbalance between empty gradients across ranks. Example: rank 0 has 2 parameters with gradients and rank 1 has 1 parameter with gradients and 1 parameter where grads are None. This caused the all-reduce to be imbalanced in size since we we're previously ignoring all grads that were None. Instead now we pad all None gradients with zero tensors. Unfortunately this imbalance did not cause a crash it would cause hanging issues in the first place that tried to access the reduced gradient data itself.