Add per-factor eigenvalue correction for Distributed Shampoo#263
Open
runame wants to merge 6 commits into
Open
Add per-factor eigenvalue correction for Distributed Shampoo#263runame wants to merge 6 commits into
runame wants to merge 6 commits into
Conversation
…rsUnwrapped base class Consolidate duplicated eigendecomposition logic from EigendecomposedShampooKroneckerFactorsUnwrapped and EigenvalueCorrectedShampooKroneckerFactorsUnwrapped into a shared base class. The base class provides _perform_eigendecomposition and _amortized_computation, with subclass behavior controlled via hasattr checks on field presence. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…or_outer_product hook Inline the outer product loop into BaseShampooPreconditionerList._update_factor_matrices and introduce _transform_grad_for_outer_product as the single extension point. The base returns grad unchanged; KL-Shampoo subclasses override it to precondition the gradient. This eliminates _compute_outer_product_list from all three classes that defined it. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Introduce PerFactorEigenvalueCorrectedShampoo, which stores m+n eigenvalues per block (one per factor dimension) computed directly as diag(Q^T M Q), where Q are cached eigenvectors and M is the already-accumulated factor matrix. This is more memory-efficient than EShampoo/SOAP's m*n eigenvalues while still providing eigenvalue correction. New classes: - PerFactorEigenvalueCorrectedShampooKroneckerFactorsUnwrapped - PerFactorEigenvalueCorrectedShampooPreconditionerList - PerFactorEigenvalueCorrectedKLShampooPreconditionerList (KL variant) - PerFactorEigenvalueCorrectedShampooPreconditionerConfig - PerFactorEigenvalueCorrectedKLShampooPreconditionerConfig Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Test the combined PerFactor+KL variant which recomputes eigenvalues every step and preconditions gradients before outer products. Uses beta2=0 and epsilon=1.0 to get clean expected values, leveraging the perturb_before_computation happy path where KL is effectively a no-op when eigenvalues are equal. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…sses Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
3 tasks
meta-cla
Bot
added
the
CLA Signed
|
@hjmshi has imported this pull request. If you are a Meta employee, you can view this in D104875458. |
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.
Summary
PerFactorEigenvalueCorrectedShampooPreconditionerListand its KL variant. These store eigenvectors and eigenvalues per factor matrix likeEigendecomposedShampoo, but recompute eigenvalues every iteration asdiag(Q^T M Q)instead of from eigendecomposition. Eigenvectors are still updated via amortized eigendecomposition.PerFactorEigenvalueCorrectedShampooPreconditionerConfigandPerFactorEigenvalueCorrectedKLShampooPreconditionerConfigtoshampoo_types.pyand export them fromdistributed_shampoo/__init__.py.epsilon=1.0for clean expected values.Stack
This PR is part of a stack adding per-factor eigenvalue correction to Distributed Shampoo:
_transform_grad_for_outer_producthook (KL refactor)Test plan
PerFactorEigenvalueCorrectedShampooPreconditionerListpassPerFactorEigenvalueCorrectedKLShampooPreconditionerListpassdistributed_shampoo/tests/,distributed_shampoo/preconditioner/tests/)make type-check)Generated with Claude Code