GRACE: Reinforcement Learning for Grounded Response and Abstention under Contextual Evidence

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhao, Yibo, Zhu, Jiapeng, Ding, Zichen, Li, Xiang
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918277816516608
author Zhao, Yibo
Zhu, Jiapeng
Ding, Zichen
Li, Xiang
author_facet Zhao, Yibo
Zhu, Jiapeng
Ding, Zichen
Li, Xiang
contents 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..
format Preprint
id arxiv_https___arxiv_org_abs_2601_04525
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle GRACE: Reinforcement Learning for Grounded Response and Abstention under Contextual Evidence
Zhao, Yibo
Zhu, Jiapeng
Ding, Zichen
Li, Xiang
Computation and Language
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..
title GRACE: Reinforcement Learning for Grounded Response and Abstention under Contextual Evidence
topic Computation and Language
url https://arxiv.org/abs/2601.04525