跳到正文
返回论文列表
cs.LG提交于 已译

大语言模型中双重用途概念的正交边界遗忘

Orthogonal Boundary Unlearning for Dual-Use Concepts in Large Language Models

Niccol\`o Ronchetti

中文摘要

为消除大语言模型在网络安全与病原体增强等高风险领域的危险能力,同时保留良性科学效用,需解决双重用途领域中合法防御与恶意攻击共享词汇及内部表征导致的标准遗忘方法严重副作用(概念泄漏)问题。正交边界遗忘(OBU)框架通过奇异值分解从校准查询中提取领域匹配的能力子空间,并将参数更新严格限制在正交零空间内,确保良性表征在校准子空间内零扰动。OBU 包含两种实现:在迭代梯度微调中应用正交约束的投影 LoRA,以及无需训练的单步解析权重手术。在 Gemma-3-4B、Qwen3.5-4B、SmolLM3-3B 和 Granite-4.2-3B 四个开源模型上,OBU 实现了双重用途概念的有效分离,显著缓解了上下文提示攻击,并在不降低良性领域能力的前提下展现了对对抗性再学习的持久抵抗力。

关键要点

  1. 01解决双重用途领域中良性防御与恶意攻击共享表征导致标准遗忘方法产生严重概念泄漏的问题。
  2. 02提出正交边界遗忘框架,利用奇异值分解提取能力子空间,将参数更新限制在正交零空间内。
  3. 03提供投影 LoRA 和闭式权重手术两种实现,分别适用于迭代微调与单步解析权重编辑。
  4. 04在四个开源基础模型上验证,OBU 有效分离双重用途概念并抵御对抗性再学习,且不损害良性能力。

解读

尚无解读。

原始英文摘要

arXiv:2610.27490v1 Announce Type: new Abstract: Mitigating catastrophic risks from large language models requires excising hazardous operational capabilities, such as cyberwarfare exploitation and pathogen enhancement, without degrading benign scientific utility. In these dual-use domains, legitimate defensive science and malicious attacks share vocabulary, syntax, and internal model representations, causing standard unlearning methods to induce severe collateral damage (concept bleeding). In this work, we present Orthogonal Boundary Unlearning (OBU), a mathematically grounded framework designed to eliminate collateral damage on benign dual-use capabilities. OBU extracts a domain-matched capability subspace from paired calibration queries via singular value decomposition on empirical activation covariances, and restricts parameter updates strictly to the orthogonal null space. This guarantees that benign representations experience zero parameter perturbation along the spanned calibration subspace. We present two complementary realizations of OBU: Projected LoRA, which applies the orthogonal constraint during iterative gradient-based fine-tuning, and Closed-Form Weight Surgery, which directly edits model weights in a single analytical step without training. Across four modern open-weight foundation models (Gemma-3-4B, Qwen3.5-4B, SmolLM3-3B, and Granite-4.2-3B), OBU achieves strong dual-use concept separation, substantially mitigates in-context priming attacks, and demonstrates durable resistance against adversarial relearning without degrading benign domain competence.

同方向论文 · cs.LG

查看全部 →

反向发现 · 这篇论文相关的工具与提示词

去交叉发现 →

关联工具 · 1

The Decoder85

Anthropic 发布 Claude Fable 5.2 与 Mythos 5.2,IPO 前重回基准榜首

Anthropic 发布 Claude Fable 5.2 与 Mythos 5.2,两者共享同一模型底座,区别在于访问权限:Fable 面向所有用户并屏蔽网络安全和生物领域的风险任务,Mythos 解除部分限制,仅向通过 Anthropic 网络安全或生命科学验证计划的组织开放。

推荐理由:材料给出 Fable 5.2 与 Mythos 5.2 的基准排名、定价和受限开放方式,可对照其与 Opus 5.5 的取舍。