Simplify the SplitState application for optimizers TBE SSD #4492
Conversation
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
Summary: Currently, the split state application for optimizer states in TBE SSD presumes Exact Rowwise Adagrad, which has only one optimizer state. This change extends the split application to support optimizers with more than one state, such as Partial Rowwise Adam. The code draws inspiration from `construct_split_state()` in non-SSD TBE, but is much more simplified and declarative and thus ergonomic to use Reviewed By: sryap, emlin, ionuthristodorescu Differential Revision: D76709101
✅ Deploy Preview for pytorch-fbgemm-docs ready!
To edit notification comments on pull requests, go to your Netlify project configuration. |
|
This pull request was exported from Phabricator. Differential Revision: D76709101 |
|
This pull request has been merged in 16cf235. |
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Outdated suggestions cannot be applied.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.
Summary:
Currently, the split state application for optimizer states in TBE SSD presumes Exact Rowwise Adagrad, which has only one optimizer state. This change extends the split application to support optimizers with more than one state, such as Partial Rowwise Adam.
The code draws inspiration from
construct_split_state()in non-SSD TBE, but is much more simplified and declarative and thus ergonomic to useReviewed By: sryap, emlin, ionuthristodorescu
Differential Revision: D76709101