Keep failed ROAD safe merges from changing base layers - #3691
Open
Robertboy18 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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
I noticed the same safe-merge ordering issue fixed for LoRA in #3611 also exists in ROAD. The regular layer writes the transformed weight before checking the transformed bias, while the 4-bit and 8-bit paths don't check the transformed bias at all.
This computes and validates both candidates before committing either one, and adds regression coverage for the regular, 4-bit, and 8-bit paths.
I ran the full ROAD test subset, all seven ROAD GPU tests on an A100, and the Ruff formatting/lint checks :)