通过删除单一向量，轻松解决大模型误拒安全请求的问题。
Surgical, Cheap, and Flexible: Mitigating False Refusal in Language Models via Single Vector Ablation
- 提取导致误拒的特定向量并直接删除
- 误拒率显著下降，同时保持安全性和通用能力
- 无需训练、适配任意模型，适合实际部署
训练语言模型做到既有益又无害，需精细调节其拒绝行为：模型应拒绝恶意指令（如“如何杀人？”），但不应拒绝看似相似却安全的请求（如“如何终止一个Python进程？”）。已有研究显示，即使高性能模型也难以避免此类误拒。本文提出一种简单且精准的方法——单向量消融，针对特定模型提取导致误拒的向量，并证明将其移除可有效降低误拒率，同时保持模型的安全性与通用能力。该方法无需训练、适用于任何模型，可用于对模型安全性进行细粒度调控，为当前及未来语言模型缓解误拒问题提供实用方案。
原文摘要 · Abstract (English)
Training a language model to be both helpful and harmless requires careful calibration of refusal behaviours: Models should refuse to follow malicious instructions or give harmful advice (e.g."how do I kill someone?"), but they should not refuse safe requests, even if they superficially resemble unsafe ones (e.g. "how do I kill a Python process?"). Avoiding such false refusal, as prior work has shown, is challenging even for highly-capable language models. In this paper, we propose a simple and surgical method for mitigating false refusal in language models via single vector ablation. For a given model, we extract a false refusal vector and show that ablating this vector reduces false refusal rate while preserving the model's safety and general capabilities. We also show that our approach can be used for fine-grained calibration of model safety. Our approach is training-free and model-agnostic, making it useful for mitigating the problem of false refusal in current and future language models.
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。