docs: Plan für Claim-level Faithfulness-Gate#68
Merged
Conversation
Cross-Model-ueberarbeiteter Plan (Codex + Mistral) fuer ein nachgelagertes Gate, das Fehlattribution + Detail-Extrapolation faengt, die der anker-basierte Verifier strukturell uebersieht (Hrastinski-Lauf 2026-06-24). M2-Roadmap.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Persistiert den cross-model-überarbeiteten Plan (
internal/docs/faithfulness-gate-plan.md) für ein nachgelagertes Faithfulness-Gate.Motivation: Der Hrastinski-Lauf (2026-06-24) erzeugte eine Fehlattribution + eine Detail-Extrapolation, die der anker-basierte Verifier strukturell übersieht (er prüft nur Zitat-Existenz auf Seitenebene, nicht Detail-Treue der Paraphrase). Pipeline-Eval meldete hall=0,000; erst ein Qwen-Faktencheck fing die Fehler.
Plan-Kern: Page-Index → High-Risk-Claim-Dekomposition → Attribution-Heuristik + mDeBERTa/XNLI-Entailment → any-high-risk-claim-Hard-Gate, nach finalem Critic + im Refine-Pfad. M2-Kaliber.
Cross-Model-reviewt (Codex + Mistral): beide "tragfähig, kein Pivot", 8 Findings eingearbeitet. Nur Plan-Dokument, kein Code.
🤖 Generated with Claude Code