Dion2 with shard-independent update route added#25
Open
kwangjunahn wants to merge 1 commit intomicrosoft:mainfrom
Open
Dion2 with shard-independent update route added#25kwangjunahn wants to merge 1 commit intomicrosoft:mainfrom
kwangjunahn wants to merge 1 commit intomicrosoft:mainfrom
Conversation
|
@kwangjunahn please read the following Contributor License Agreement(CLA). If you agree with the CLA, please reply with the following information.
Contributor License AgreementContribution License AgreementThis Contribution License Agreement (“Agreement”) is agreed to by the party signing below (“You”),
|
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This PR adds a shard_independent argument to the Dion2 optimizer, giving users control over how top-K row selection interacts with FSDP sharding.
1. Sharding-independent row selection for FSDP
Dion2 selects a subset of rows (top-K by L1 norm) for orthogonalization at each step. When parameters are sharded across devices, there's a design choice: should selection happen on local shards or on the full matrix?
Options
shard_independent=False(default)shard_independent=True2. Row-wise Selection Only
We now restrict Dion2's selection to row-wise selection (i.e.,
select_dim=-2). This ensures the selection semantics match "neuron-wise" selection, since linear layers have shape (output, input) — each row corresponds to one output neuron.Previously, the code would choose between row or column selection based on matrix shape or sharding configuration. This change makes the behavior more interpretable and consistent.