Fixed staff detection errors#122
Draft
liebharc wants to merge 5 commits into
Draft
Conversation
liebharc
force-pushed
the
feature/staff-detection
branch
from
July 15, 2026 16:13
f5f46d4 to
f68f8a3
Compare
Collaborator
|
Wow, this improvement looks significant! |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Fixes some staff detection problems which came up in the smb benchmark.
Disclaimer: Claude code changes which mostly accumulated while I was running it in the background, still have to review this myself
Performance changes
It turns out that SMB is also unreliable. For some pieces it omits staccato information which penalizes any model which correctly detect them. I decided to just keep it like that as it's likely not meaningful to spend the time on that.