[Train] Simplify single worker training #19814
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
matthewdeng
reviewed
Oct 28, 2021
matthewdeng
approved these changes
Oct 28, 2021
There was a problem hiding this comment.
This is awesome. Really like this pattern of running the documentation code in CI.
Can you add instructions and link this an example in [OSS] How to write an example? This is really a best-practice I think we should all be following. (Also be sure to include the trick of moving the function calls to __main__!)
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Outdated suggestions cannot be applied.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.
Currently, Ray Train does not setup the distributed environment (torch process group, TF_CONFIG env var) if only using 1 worker.
However, this requires the user to make changes to their training code if they want to go from 1 worker to multiple workers, and has been a source of confusion in our examples:
#19506
#19761
This PR changes the behavior to setup the distributed environment regardless of the number of workers. This allows training functions that have
DistributedDataParallelorMultiWorkerMirroredStrategyto still work with single worker Ray Train. This PR also adds testing the quick start code examples in the docs.Closes #19761
Why are these changes needed?
Related issue number
Checks
scripts/format.shto lint the changes in this PR.