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 2404.11972cs.CL
收藏

使语言模型对齐以显式处理歧义

Aligning Language Models to Explicitly Handle Ambiguity

Seoul National University(首尔大学);NAVER Cloud(NAVER云);KAIST AI(KAIST人工智能);NAVER AI LAB(NAVER人工智能实验室);Hanyang University(汉阳大学)

Hyuhng Joon Kim, Youna Kim, Cheonbok Park, Junyeob Kim, Choonghyun Park, Kang Min Yoo, Sang-goo Lee, Taeuk Kim

AI总结:

针对大语言模型未显式训练处理歧义、感知歧义程度受知识影响的问题,提出感知歧义对齐(APA)流程,利用模型自身歧义评估对齐处理歧义查询,实验显示其效果优于金标准标签训练,尤其在分布外场景表现更优。

摘要:

在用户与语言模型智能体的交互中,用户话语常为提升效率而出现省略(遗漏单词或短语)或表述不精确(缺乏确切性)的情况。这可能导致同一输入因不同假设或背景知识而产生不同解读。因此,智能体必须熟练处理查询中固有的歧义,以确保可靠性。然而,即便是最先进的大语言模型(LLMs)在此类场景下仍面临挑战,主要障碍包括:(1)LLMs未经过处理歧义话语的显式训练;(2)LLMs感知到的歧义程度可能因自身掌握的知识而有所不同。为解决这些问题,我们提出了感知歧义对齐(Alignment with Perceived Ambiguity,APA),这是一种新型流程,通过利用LLMs自身对歧义的评估(即感知歧义),使LLMs对齐以处理歧义查询。在问答数据集上的实验结果表明,APA能让LLMs显式检测并处理歧义查询,同时保留回答清晰问题的能力。此外,我们的研究结果证明,APA的表现优于使用金标准标签的训练方式,尤其在分布外场景中效果更突出。数据和代码可在https://github.com/heyjoonkim/APA获取。

英文摘要:

In interactions between users and language model agents, user utterances frequently exhibit ellipsis (omission of words or phrases) or imprecision (lack of exactness) to prioritize efficiency. This can lead to varying interpretations of the same input based on different assumptions or background knowledge. It is thus crucial for agents to adeptly handle the inherent ambiguity in queries to ensure reliability. However, even state-of-the-art large language models (LLMs) still face challenges in such scenarios, primarily due to the following hurdles: (1) LLMs are not explicitly trained to deal with ambiguous utterances; (2) the degree of ambiguity perceived by the LLMs may vary depending on the possessed knowledge. To address these issues, we propose Alignment with Perceived Ambiguity (APA), a novel pipeline that aligns LLMs to manage ambiguous queries by leveraging their own assessment of ambiguity (i.e., perceived ambiguity). Experimental results on question-answering datasets demonstrate that APA empowers LLMs to explicitly detect and manage ambiguous queries while retaining the ability to answer clear questions. Furthermore, our finding proves that APA excels beyond training with gold-standard labels, especially in out-of-distribution scenarios. The data and code are available at https://github.com/heyjoonkim/APA.

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。

↑