GRACE: Reinforcement Learning for Grounded Response and Abstention under Contextual Evidence
This addresses a critical reliability issue in retrieval-augmented generation for large language models, offering a unified solution with practical efficiency gains.
The paper tackles the problem of retrieval-augmented generation systems providing ungrounded correct answers or fabricated responses when evidence is insufficient, proposing GRACE, a reinforcement learning framework that achieves state-of-the-art overall accuracy and balances accurate response with rejection while reducing annotation costs by 90%.
Retrieval-Augmented Generation (RAG) integrates external knowledge to enhance Large Language Models (LLMs), yet systems remain susceptible to two critical flaws: providing correct answers without explicit grounded evidence and producing fabricated responses when the retrieved context is insufficient. While prior research has addressed these issues independently, a unified framework that integrates evidence-based grounding and reliable abstention is currently lacking. In this paper, we propose GRACE, a reinforcement-learning framework that simultaneously mitigates both types of flaws. GRACE employs a data construction method that utilizes heterogeneous retrievers to generate diverse training samples without manual annotation. A multi-stage gated reward function is then employed to train the model to assess evidence sufficiency, extract key supporting evidence, and provide answers or explicitly abstain. Experimental results on two benchmarks demonstrate that GRACE achieves state-of-the-art overall accuracy and strikes a favorable balance between accurate response and rejection, while requiring only 10% of the annotation costs of prior methods. Our code is available at https://github.com/YiboZhao624/Grace..