arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2406.15673cs.CLcs.AI
收藏

大型语言模型具有内在自我纠正能力

Large Language Models have Intrinsic Self-Correction Ability

University at Buffalo(布法罗大学);University of Notre Dame(圣母大学)

Dancheng Liu, Amir Nassereldine, Ziming Yang, Chenhui Xu, Yuting Hu, Jiajie Li, Utkarsh Kumar, Changjae Lee, Ruiyang Qin, Yiyu Shi, Jinjun Xiong

AI总结:

本文通过理论分析和实证实验,证明大型语言模型具备内在自我纠正能力,并发现零温度和公平提示是实现该能力的两个关键因素。

摘要:

大型语言模型(LLMs)因其在各种自然语言处理任务中的卓越能力而备受关注,但它们存在会导致性能下降的幻觉问题。改进LLMs性能的一个有前景的解决方案是要求LLMs在生成后修改其答案,这种技术被称为自我纠正。在两种类型的自我纠正中,内在自我纠正被认为是一个有前景的方向,因为它不利用外部知识。然而,最近的研究对LLM进行内在自我纠正的能力的有效性提出了质疑。在本文中,我们通过理论分析和实证实验,对LLMs的内在自我纠正能力提出了一个新的视角。此外,我们确定了成功自我纠正的两个关键因素:零温度和公平的提示。利用这些因素,我们证明了内在自我纠正能力在多个现有的LLMs中得到了体现。我们的发现为LLMs自我纠正行为背后的基本理论提供了见解,并强调了无偏提示和零温度设置在发挥其全部潜力方面的重要性。

英文摘要:

Large language models (LLMs) have attracted significant attention for their exceptional abilities in various natural language processing tasks, but they suffer from hallucinations that will cause performance degradation. One promising solution to improve the LLMs' performance is to ask LLMs to revise their answer after generation, a technique known as self-correction. Among the two types of self-correction, intrinsic self-correction is considered a promising direction because it does not utilize external knowledge. However, recent works doubt the validity of LLM's ability to conduct intrinsic self-correction. In this paper, we present a novel perspective on the intrinsic self-correction capabilities of LLMs through theoretical analyses and empirical experiments. In addition, we identify two critical factors for successful self-correction: zero temperature and fair prompts. Leveraging these factors, we demonstrate that intrinsic self-correction ability is exhibited across multiple existing LLMs. Our findings offer insights into the fundamental theories underlying the self-correction behavior of LLMs and remark on the importance of unbiased prompts and zero temperature settings in harnessing their full potential.

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。

↑