Regime-Aware Peer Specialization for Robust RAG under Heterogeneous Knowledge Conflicts

2026-06-29

AI Summary

This work addresses the vulnerability of Retrieval-Augmented Generation (RAG) when external context conflicts with internal knowledge, particularly in heterogeneous reliability settings where distinguishing valid learning signals is challenging. The authors propose RAPS-DA, a novel framework that decouples such conflicts at the sample level into three mechanisms—Grounding, Arbitration, and Resistance—and trains same-size expert models routed via hard assignment. At the token level, a dual-layer selector dynamically adjusts supervision emphasis by integrating teacher disagreement, teacher–student divergence, and student entropy. RAPS-DA introduces the first mechanism-aware expert specialization strategy, enabling fine-grained modeling under fixed model capacity without requiring mechanism labels or expert models during inference. Experiments demonstrate that RAPS-DA significantly outperforms existing baselines across five conflict scenarios and two out-of-distribution benchmarks.

Abstract

Retrieval-augmented generation (RAG) improves language models by grounding generation in external context. However, it can be fragile when the retrieved context conflicts with the model's parametric knowledge. Such conflicts span a reliability spectrum, ranging from reliable and partially reliable evidence to adversarial context. Existing remedies often handle such heterogeneous conflicts with regime-agnostic supervision, which can conflate incompatible learning signals across reliability regimes. To disentangle these signals, we propose RAPS-DA, a regime-aware peer specialization framework that addresses conflict at two complementary granularities. At the sample level, conflicts are divided into three regimes, including Grounding, Arbitration, and Resistance, with one same-scale peer specialist trained per regime from a shared base model. Each sample is then hard-routed to its regime-matched peer for on-policy reverse-KL supervision. At the token level, a dual-layer selector uses inter-teacher disagreement, student-teacher divergence, and student entropy to filter uninformative or unstable tokens, upweight confidently misaligned ones, and gradually focus supervision on high-conflict tokens as the student matures. Gains stem from specialization at a fixed model scale, not from a stronger teacher, and the peer specialists exist only during training, so the deployed student requires no regime labels or peer access. Experiments on five conflict scenarios and two out-of-distribution benchmarks show RAPS-DA surpasses all prompting, decoding, fine-tuning, RL, and single-teacher baselines.

Problem

• Retrieval-Augmented Generation
• Knowledge Conflict
• Heterogeneous Reliability
• Regime-Aware Learning
• Robustness

Innovation

• Regime-Aware Learning
• Peer Specialization
• Retrieval-Augmented Generation
• Knowledge Conflict Resolution
• Token-Level Supervision