Fix of SwiGLU hidden not being multiple of world size #390
Conversation
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
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Outdated suggestions cannot be applied.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.
What does this PR do?
This PR fixe an issue that occurs when using SwiGLU in a FSDP2 + TP combination.
Since the hidden dim is computed dynamically to compensate for the difference in the number of trainable weights in comparison to GeLU, we end up with hidden dims that are not a multiple of the world size.
This in return creates a problem when applying TP and subsequently FSDP2, as uneven sharding is not allowed for FSDP2 + TP (yet). The issue does not occur when only using FSDP2, as FSDP2 already support uneven sharding out of the box.
General Changes
Breaking Changes
Checklist before submitting final PR
python tests/tests.py)CHANGELOG_DEV.md)