arXiv 2410.123272026-06-11cs.CL

基于神经元的大语言模型人格特质诱导

Neuron-based Personality Trait Induction in Large Language Models

Gaoling School of Artificial Intelligence, Renmin University of China（中国人民大学首都人工智能学院）；Tongyi Lab（通义实验室）；Institute of Statistics and Big Data, Renmin University of China（中国人民大学统计与大数据研究院）

Jia Deng, Tianyi Tang, Yanbin Yin, Wenhao Yang, Wayne Xin Zhao, Ji-Rong Wen

备注：: 25 pages. Published at ICLR 2025
发表信息：: The Thirteenth International Conference on Learning Representations (ICLR 2025), 2025

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

提出基于神经元的大语言模型人格特质诱导方法，通过构建PersonalityBench数据集、识别人格相关神经元并调整其值，实现无需训练的参数级控制，效果媲美微调模型。

摘要：

大型语言模型（LLMs）在模拟各种人格特质方面变得越来越熟练，这是支持相关应用（例如角色扮演）的重要能力。为了进一步提高这种能力，在本文中，我们提出了一种基于神经元的方法，用于LLMs的人格特质诱导，包含三个主要技术贡献。首先，我们构建了PersonalityBench，一个用于识别和评估LLMs人格特质的大规模数据集。该数据集基于心理学中的大五人格特质，旨在评估LLMs对特定人格特质的生成能力。其次，通过利用PersonalityBench，我们提出了一种高效的方法，通过检查给定特质的对立面来识别LLMs中与人格相关的神经元。第三，我们开发了一种简单而有效的诱导方法，通过操纵这些已识别的人格相关神经元的值。该方法无需训练和修改模型参数即可实现对LLMs所表现特质的细粒度控制。大量实验验证了我们神经元识别和特质诱导方法的有效性。值得注意的是，我们的方法实现了与微调模型相当的性能，为LLMs的人格特质诱导提供了更高效、更灵活的解决方案。我们在以下网址提供所有提到的资源：此 https URL。

英文摘要：

Large language models (LLMs) have become increasingly proficient at simulating various personality traits, an important capability for supporting related applications (e.g., role-playing). To further improve this capacity, in this paper, we present a neuron-based approach for personality trait induction in LLMs, with three major technical contributions. First, we construct PersonalityBench, a large-scale dataset for identifying and evaluating personality traits in LLMs. This dataset is grounded in the Big Five personality traits from psychology and is designed to assess the generative capabilities of LLMs towards specific personality traits. Second, by leveraging PersonalityBench, we propose an efficient method for identifying personality-related neurons within LLMs by examining the opposite aspects of a given trait. Third, we develop a simple yet effective induction method that manipulates the values of these identified personality-related neurons. This method enables fine-grained control over the traits exhibited by LLMs without training and modifying model parameters. Extensive experiments validate the efficacy of our neuron identification and trait induction methods. Notably, our approach achieves comparable performance as fine-tuned models, offering a more efficient and flexible solution for personality trait induction in LLMs. We provide access to all the mentioned resources at https://github.com/RUCAIBox/NPTI.

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。