AffineTransform: handle empty MLX input in block-diagonal path - #183
Merged
Conversation
MLX's scatter (indexed assignment) rejects zero-size updates, so result[..., out_idx] = matmul(...) raised e.g. 'ValueError: [scatter] Updates with shape (0,32,1,1) are too large for array with shape (0,128)' when an empty startup message reached a block-diagonal transform. The pre-allocated zeros output is already the correct result for any zero-length input dimension (a zero-length inner dimension sums to zero), so return it without entering the cluster-assignment loop. The emptiness check reads only array metadata, so no backend conversion or evaluation occurs. Also includes a ruff-format cleanup of an adjacent f-string.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
Fix
AffineTransformTransformerwhen an empty message is processed through the block-diagonal MLX path.MLX rejects indexed scatter assignments with zero-size updates. The block-diagonal implementation previously attempted one such assignment per cluster even though the preallocated zero-size output was already complete, causing empty chunks to raise.
Changes
Testing
tests/unit/test_affine_transform.py: 47 passed.