destroy process group in end_training
#3012
Merged
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
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
|
All of the trackers come loaded with |
muellerzr
reviewed
Aug 15, 2024
| @@ -2678,11 +2678,10 @@ def log(self, values: dict, step: int | None = None, log_kwargs: dict | None = { | |||
| for tracker in self.trackers: | |||
| tracker.log(values, step=step, **log_kwargs.get(tracker.name, {})) | |||
|
|
|||
| @on_main_process | |||
| def end_training(self): | |||
There was a problem hiding this comment.
Note that eventually we'll get to a point where users will want to pass in their own pg's and init them, e.g. this is needed for hybrid shard w/ FSDP. So we may need to take that in there potentially. Will see how it ends up looking
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Outdated suggestions cannot be applied.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 ?
With the latest version of torch, when we use multi-gpu, torch will trigger a warning asking us to call
destoy_process_group(). This PR fixes this by adding that in end_training method. Note that this function needs to be called on all process.For trackers, we are already only executing the methods on the main process. See here. So it should be safe to remove the on_main_process decorator in the end_training method.
However, I see for WandBTracker that
main_process_only = False. Is there a specific reason @muellerzr ?