eval(no-count-drift): committed recall probe (in-scope coverage 25/25)#28
Merged
Conversation
… 25/25) Answers the open question from the v6 merge: precision was proven (0/988 independent FP) but recall was unmeasured because the MAD corpus barely contains the target pattern. Adds evaluation/v6/recall_probe.jsonl — 25 genuine count-drift positives authored to span phrasing variety (digit/word lead-ins, number-first headings, prose prefixes, "all N passed", "there/here are N", numbered lists, "a dozen", N-of-M, fraction/percent) — and folds recall into score_count_drift.py / RESULTS.md. Result: 25/25 in-scope recall, alongside 0/988 independent FP. Conclusion (data-backed): no LLM-judge tier or Rust engine is warranted — the deterministic tier already covers the common phrasing space at full precision. Out-of-scope forms remain abstained by design. No hook logic changed. Co-Authored-By: Claude Opus 4.7 (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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Follow-up to #27. Adds a committed recall probe (
evaluation/v6/recall_probe.jsonl, 25 genuine count-drift positives spanning phrasing variety) and folds recall into the scorer/RESULTS. Result: 25/25 in-scope recall alongside the 0/988 independent false positives from #27. Data-backed conclusion: no LLM-judge tier or engine is warranted — the deterministic tier covers the common phrasing space at full precision; out-of-scope forms stay abstained by design. No hook logic changed.🤖 Generated with Claude Code