Add generator selection-bias analysis for the perturbation benchmark#100
Open
dangng2004 wants to merge 1 commit into
Open
Add generator selection-bias analysis for the perturbation benchmark#100dangng2004 wants to merge 1 commit into
dangng2004 wants to merge 1 commit into
Conversation
Compares the feature distribution of perturbation spans the generator selected against the full candidate pool it chose from. Random selection would make the two distributions match, so divergence measures selection bias directly, with no new LLM calls. Walks existing *_perturbations.json files and re-extracts the candidate pool with the same extractor. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Quantifies selection bias in the perturbation generator: the LLM picks which candidate spans to perturb, so a selected subset whose feature distribution diverges from the candidate pool measures bias directly, with no new LLM calls.
Walks existing
*_perturbations.jsonfiles, re-extracts the candidate pool with the same extractor (extract.py), and compares selected vs pool distributions.🤖 Generated with Claude Code