[Train] Torch data transfer automatic conversion #20333
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
Co-authored-by: matthewdeng <[email protected]>
Co-authored-by: matthewdeng <[email protected]>
matthewdeng
reviewed
Nov 14, 2021
amogkam
changed the title
save_checkpoint automatic conversion
matthewdeng
approved these changes
Nov 14, 2021
python/ray/train/backend.py
Outdated
| @@ -66,13 +122,14 @@ class BackendExecutor: | |||
| def __init__( | |||
| self, | |||
| backend_config: BackendConfig, | |||
| backend: Backend, | |||
There was a problem hiding this comment.
nit: Undo this change since it doesn't make sense to have conflicting BackendConfig and Backend.
There was a problem hiding this comment.
Ok I made Backend into a singleton, so it's ok to instantiate any of the backends any number of times, and we don't need to pass around a single instance.
6 tasks
Closed
2 tasks
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Outdated suggestions cannot be applied.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.
When saving a model in a checkpoint or reporting a model, the user has to manually extract out the module from DDP and move it to cpu so that it can be properly deserialized on the driver.
This PR adds functionality to automatically do the above so the user does not have to add this logic to their training script.
TODO: Possibly use this logic not just for checkpoints, but for return values as well.
Why are these changes needed?
Related issue number
Checks
scripts/format.shto lint the changes in this PR.