用少量计算修复低资源语言的模型安全漏洞
Multilingual Safety Alignment Via Sparse Weight Editing
- 通过稀疏权重编辑定位安全神经元，实现跨语言对齐
- 在8种语言上将攻击成功率大幅降低，推理能力基本不变
- 无需训练、仅需一次计算，适合资源有限的场景
大型语言模型在不同语言间存在显著安全差异，低资源语言（LRLs）常绕过为高资源语言（HRLs）如英语设定的安全防护。现有方法如多语言监督微调（SFT）或基于人类反馈的强化学习（RLHF）成本高且依赖稀缺的多语言安全数据。本文提出一种无需训练的对齐框架——稀疏权重编辑。我们发现安全能力集中于少量安全神经元，将跨语言对齐建模为带约束的线性变换，推导出闭式解，可最优地将LRL的有害表征映射至HRL的鲁棒安全子空间，同时通过零空间投影约束保留通用能力。在8种语言和多个模型族（Llama-3、Qwen-2.5）上的实验表明，该方法在单次高效计算下显著降低LRL的攻击成功率（ASR），对通用推理能力影响极小。
原文摘要 · Abstract (English)
Large Language Models (LLMs) exhibit significant safety disparities across languages, with low-resource languages (LRLs) often bypassing safety guardrails established for high-resource languages (HRLs) like English. Existing solutions, such as multilingual supervised fine-tuning (SFT) or Reinforcement Learning from Human Feedback (RLHF), are computationally expensive and dependent on scarce multilingual safety data. In this work, we propose a novel, training-free alignment framework based on Sparse Weight Editing. Identifying that safety capabilities are localized within a sparse set of safety neurons, we formulate the cross-lingual alignment problem as a constrained linear transformation. We derive a closed-form solution to optimally map the harmful representations of LRLs to the robust safety subspaces of HRLs, while preserving general utility via a null-space projection constraint. Extensive experiments across 8 languages and multiple model families (Llama-3, Qwen-2.5) demonstrate that our method substantially reduces Attack Success Rate (ASR) in LRLs with negligible impact on general reasoning capabilities, all achieved with a single, data-efficient calculation.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。