Beyond Black-Box Interventions: Latent Probing for Faithful Retrieval-Augmented Generation

ArXi:2510.12460v2 Announce Type: replace Retrieval-Augmented Generation (RAG) systems often fail to maintain contextual faithfulness, generating responses that conflict with the provided context or fail to fully leverage the provided evidence. Existing methods attempt to improve faithfulness through external interventions, such as specialized prompting, decoding-based calibration, or preference optimization. However, since these approaches treat the LLM as a black box, they lack a reliable mechanism to assess when and why knowledge conflicts occur.