Fix matmul_4bit gemv path for mismatched quant_state layout - #1906
Closed
datavorous wants to merge 2 commits into
Closed
Fix matmul_4bit gemv path for mismatched quant_state layout#1906datavorous wants to merge 2 commits into
datavorous wants to merge 2 commits into
Conversation
Copy link
Copy Markdown
Sorry, something went wrong.
Copy link
Copy Markdown
Member
|
Thanks for the fix. This is superseded by #1949 which handles weight orientation more comprehensively - matmul_4bit now normalizes packed weights to a canonical shape internally and explicitly detects the transposed (in_features, out_features) layout, emitting a DeprecationWarning and falling back correctly. Since we are deprecating support for that layout, silently guarding it would go the wrong direction. Closing in favor of #1949. |
Sorry, something went wrong.
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
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Fixes #1862
Problem: gemv fast path in matmul_4bit assumes quant_state.shape follows (out_features, in_features), and can silently produce wrong output shape/values for transposed layout metadata.
Fix: add a minimal shape guard in matmul_4bit vector fast path; on mismatch, fall back to MatMul4Bit.apply.
Scope: no kernel changes, no API changes, minimal 2-line guard.