CLEAR：用於效用保持之LLM安全對齊的連續潛在適配器路由

摘要

提升大型語言模型（LLMs）的安全性往往以犧牲效用為代價，因為全域套用的安全調校可能會影響模型對有害與良性輸入的反應。我們提出連續潛在適配器路由（Continuous LatEnt Adapter Routing, CLEAR），這是一個條件式安全調適框架，利用輕量級隱藏狀態門控來持續控制安全低秩適配器的激活強度。CLEAR旨在減少有害的生成內容，同時避免對凍結的主幹模型進行不必要改變，以免降低其對良性提示詞的效能。在廣泛使用的安全性與效用基準上的實驗顯示，CLEAR提升了在HarmBench上的穩健性，同時減少了全域套用安全調校（如SFT或標準低秩適配（LoRA））所導致的效用下降。在Llama-3-8B-Instruct上，CLEAR將HarmBench的攻擊成功率（ASR）從32.3%降至0.5%，同時保留了基礎模型大部分的效用，並在GSM8K準確率上比全域套用的SFT或LoRA高出最多7.1個百分點。這些結果表明，CLEAR是改善LLM對齊中安全性—效用權衡的一種具前景的機制。

English

Improving the safety of large language models (LLMs) often comes at the expense of utility, as globally applied safety tuning may affect model responses to both harmful and benign inputs. We propose Continuous LatEnt Adapter Routing (CLEAR), a conditional safety adaptation framework that uses a lightweight hidden-state gate to continuously control the activation strength of a safety low-rank adapter. CLEAR aims to reduce harmful completions while avoiding unnecessary changes to the frozen backbone that could degrade performance on benign prompts. Experiments on widely used safety and utility benchmarks show that CLEAR improves robustness on HarmBench while reducing the utility degradation observed with globally applied safety tuning such as SFT or standard low-rank adaptation (LoRA). On Llama-3-8B-Instruct, CLEAR reduces HarmBench ASR from 32.3\% to 0.5\%, while retaining most of the base model's utility and achieving up to 7.1 percentage points higher GSM8K accuracy than globally applied SFT or LoRA. These results suggest that CLEAR is a promising mechanism for improving the safety--utility trade-off in LLM alignment.