Implement SOAP optimizer (#1580)#1617
Open
jinukuntlaakhilakumargoud-web wants to merge 2 commits intogoogle-deepmind:mainfrom
Open
Implement SOAP optimizer (#1580)#1617jinukuntlaakhilakumargoud-web wants to merge 2 commits intogoogle-deepmind:mainfrom
jinukuntlaakhilakumargoud-web wants to merge 2 commits intogoogle-deepmind:mainfrom
Conversation
Add SOAP (Second-order Optimization with Alternating Projections) optimizer to optax.contrib. SOAP improves upon Shampoo by running Adam steps in the preconditioner's eigenbasis, achieving faster convergence. - Add scale_by_soap gradient transformation and soap wrapper function - Add SoapState NamedTuple with Preconditioner for type-safe tree operations - Add helper functions for preconditioning, projection, and eigendecomposition - Add unit tests for initialization, convergence, mixed tensors, and qr_dtype - Register SOAP in contrib __init__.py and _common_test.py - Add SOAP hyperparams to uninjectable_hparam list for inject_hyperparams Reference: Vyas et al., SOAP: Improving and Stabilizing Shampoo using Adam, 2024
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Resolves #1580
Add SOAP (Second-order Optimization with Alternating Projections) optimizer to optax.contrib. SOAP improves upon Shampoo by running Adam steps in the preconditioner's eigenbasis, achieving faster convergence.
Reference: Vyas et al., SOAP: Improving and Stabilizing Shampoo using Adam, 2024