[TPU] Save checkpoint on global zero only and sync before#17882
Merged
[TPU] Save checkpoint on global zero only and sync before#17882
Conversation
Contributor
JackCaoG
approved these changes
Jun 20, 2023
|
LGTM. Calling |
Contributor
awaelchli
approved these changes
Jun 21, 2023
Borda
approved these changes
Jun 21, 2023
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
6 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What does this PR do?
#17408 changed the XLA CheckpointIO implementation to not use
xm.save().This meant that we need to save only on rank zero, but updating the logic was missed for Fabric. This bug is unreleased, so it's not mentioned in the README.
Additionally,
mark_stepis added to sync the graph before saving, to avoid any potential hangs. This is mentioned in the README as an addition.Discussed with @Liyang90
cc @Borda @carmocca @justusschock @awaelchli @JackCaoG @steventk-g @Liyang90