arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2501.13573cs.CL
收藏

通过检索头诱导优化提升大语言模型的上下文忠实性

Improving Contextual Faithfulness of Large Language Models via Retrieval Heads-Induced Optimization

Harbin Institute of Technology(哈尔滨工业大学);Peng Cheng Laboratory(鹏城实验室);Northeastern University(东北大学);The University of Hong Kong(香港大学);iFLYTEK Research(讯飞研究院)

Lei Huang, Xiaocheng Feng, Weitao Ma, Yuchun Fan, Xiachong Feng, Yangfan Ye, Weihong Zhong, Yuxuan Gu, Baoxin Wang, Dayong Wu, Guoping Hu, Bing Qin

AI总结:

本文提出RHIO框架,利用检索头相关性,通过遮蔽检索头增强不忠实样本并联合训练,结合控制标记与对比解码,显著提升长形式问答的上下文忠实性,并引入GroundBench基准验证其效果优于GPT-4o。

摘要:

在检索增强的大语言模型中,确保上下文忠实性对于构建可信的信息寻求系统至关重要,尤其是在长形式问答场景中。在本工作中,我们发现了长形式问答忠实性与检索头(一组负责检索上下文信息的注意力头)之间的显著相关性。利用这一洞见,我们提出了RHIO框架,旨在教会大语言模型明确区分忠实与不忠实的生成。RHIO首先通过选择性遮蔽检索头来增强模拟真实模型内在错误的不忠实样本。然后,将这些样本纳入联合训练,使模型能够在控制标记的条件下区分不忠实输出与忠实输出。此外,这些控制标记被用于自我诱导对比输出,通过对比解码放大它们之间的差异。为了促进上下文忠实性的评估,我们还引入了GroundBench,这是一个从五个现有长形式问答数据集汇编而成的综合基准。在GroundBench上的大量实验结果表明,RHIO显著提升了忠实性,甚至超过了GPT-4o。

英文摘要:

Ensuring contextual faithfulness in retrieval-augmented large language models (LLMs) is crucial for building trustworthy information-seeking systems, particularly in long-form question-answering (LFQA) scenarios. In this work, we identify a salient correlation between LFQA faithfulness and retrieval heads, a set of attention heads responsible for retrieving contextual information. Leveraging this insight, we propose RHIO, a framework designed to teach LLMs to explicitly discriminate between faithful and unfaithful generations. RHIO first augments unfaithful samples that simulate realistic model-intrinsic errors by selectively masking retrieval heads. Then, these samples are incorporated into joint training, enabling the model to distinguish unfaithful outputs from faithful ones conditioned on control tokens. Furthermore, these control tokens are leveraged to self-induce contrastive outputs, amplifying their difference through contrastive decoding. Additionally, to facilitate the evaluation of contextual faithfulness, we also introduce GroundBench, a comprehensive benchmark compiled from five existing LFQA datasets. Extensive experimental results on GroundBench demonstrate that RHIO significantly improves faithfulness, even outperforming GPT-4o.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑