知晓偏见，表现更好：通过Know-Bias神经元增强减轻大语言模型中的社会偏见
Knowing Bias, Doing Better: Mitigating Social Bias in LLMs via Know-Bias Neuron Enhancement
AI总结：
本文提出KnowBias框架，通过增强而非压制编码偏见知识的神经元来减轻LLM中的社会偏见，实现高效且通用的去偏方法。
摘要：
大型语言模型（LLMs）表现出的社会偏见会强化有害刻板印象，限制其安全部署。现有去偏方法多采用压制范式，通过修改参数、提示或相关神经元来抑制偏见行为，但此类方法往往脆弱、泛化能力差、数据效率低且易损害通用能力。我们提出KnowBias，一种轻量且概念上不同的框架，通过增强而非压制编码偏见知识的神经元来减轻偏见。KnowBias通过少量偏见知识问题进行属性分析，识别编码偏见知识的神经元，并在推理时选择性增强。此设计实现了强大的去偏效果，同时保持通用能力，能够跨偏见类型和人口统计数据泛化，并且数据效率高，仅需少量简单的yes/no问题且无需重新训练。在多个基准和LLM上的实验展示了稳定的当前最佳去偏性能，同时最小化了实用价值的退化。数据和代码可在https://github.com/JP-25/KnowBias获取。
英文摘要：
Large language models (LLMs) exhibit social biases that reinforce harmful stereotypes, limiting their safe deployment. Most existing debiasing methods adopt a suppressive paradigm by modifying parameters, prompts, or neurons associated with biased behavior; however, such approaches are often brittle, weakly generalizable, data-inefficient, and prone to degrading general capability. We propose \textbf{KnowBias}, a lightweight and conceptually distinct framework that mitigates bias by strengthening, rather than suppressing, neurons encoding bias-knowledge. KnowBias identifies neurons encoding bias knowledge using a small set of bias-knowledge questions via attribution-based analysis, and selectively enhances them at inference time. This design enables strong debiasing while preserving general capabilities, generalizes across bias types and demographics, and is highly data efficient, requiring only a handful of simple yes/no questions and no retraining. Experiments across multiple benchmarks and LLMs demonstrate consistent state-of-the-art debiasing performance with minimal utility degradation. Data and code are available at https://github.com/JP-25/KnowBias.
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。
