Expose apply_batch_size through InversionConfig - #450
Open
luciaquirke wants to merge 1 commit into
Open
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
EkfacConfig.apply_batch_size controls how many query-gradient columns the inverse application holds on-device at once, but the pipeline constructs EkfacConfig without it, so the default of 32 is unreachable from a run config. At 1.5B parameters with 20 per-query columns the default loads ~45 GB of query gradients on top of the eigenvectors and OOMs a 48 GB card, with no workaround short of patching the source (which is what we did in production — batching is numerically exact, so this is purely a memory knob).
This plumbs it through InversionConfig, next to the other knobs the apply stage already reads.
🤖 Generated with Claude Code
https://claude.ai/code/session_0167XKnE2yoCE162n3jHvxC1