Repository navigation
Remove the forced_choice_leakage scanner - #66
Merged
Merged
Conversation
…inary question binary_question_ratio still covers datasets dominated by binary questions. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
MattFisher
enabled auto-merge
October 8, 2026 04:50
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
Removes the
forced_choice_leakagescanner. Your call: you wondered whether it had ever been useful, and a sweep of every Inspect Evals eval found nothing to suggest it had.It flags a question that offers its answer as one of two options ("Is it an MRI or a CT scan?" → "mri"). But that's an ordinary binary question: a model that pattern-matches the phrasing still has to choose between the options. On the sweep it reported 1,614 findings:
None looks worth acting on.
binary_question_ratiostill reports datasets dominated by binary questions, which is the dataset-level form of the same concern.The scanner, its tests and its README row are removed, and PLAN.md marks it removed.
--scanners forced_choice_leakagenow fails as an unknown scanner, which the changelog fragment says.This is separate from the other sweep fixes (#62 to #65), so it can be declined without touching them.
Testing
Suite: 601 passed, excluding the browser e2e test (the scanner's own tests are gone). basedpyright: 0 errors. pre-commit passes.
🤖 Generated with Claude Code