Fix async checkpoint timing in DCP recipe#3688
Open
patrocinio wants to merge 1 commit intopytorch:mainfrom
Open
Fix async checkpoint timing in DCP recipe#3688patrocinio wants to merge 1 commit intopytorch:mainfrom
patrocinio wants to merge 1 commit intopytorch:mainfrom
Conversation
Move checkpoint_future.result() before optimizer.step() to ensure the previous checkpoint completes before weights are modified in-place. This allows better overlap of checkpointing with forward/backward passes. Fixes pytorch#3584
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/tutorials/3688
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit f914e75 with merge base 7f8b6dc ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Move checkpoint_future.result() before optimizer.step() to ensure the previous checkpoint completes before weights are modified in-place. This allows better overlap of checkpointing with forward/backward passes.
Fixes #3584
Description
Checklist
cc @wconstab @osalpekar @H-Huang @kwen2501