何时写入与何时抑制：面向记忆辅助知识编辑的路径专用双适配器
When to Write and When to Suppress: Route-Specialized Dual Adapters for Memory-Assisted Knowledge Editing
AI总结：
提出路径专用双适配器编辑器，通过相关性路由器决定是否应用编辑记忆，分别训练编辑适配器和局部性适配器，在三个基准上取得最佳概率偏好准确率。
摘要：
知识编辑系统必须更新选定的事实，同时保持邻近但无关的行为不变。本文在记忆辅助设置中研究该问题，其中在推理时检索编辑记忆，参数高效适配器校正模型的对象偏好。我们认为核心设计问题不仅是如何写入编辑，还包括何时抑制它。我们引入\method{}，一种路径专用双适配器编辑器。相关性路由器首先决定提示是否应接收编辑记忆。被路由的提示使用训练为偏好新对象而非原始对象的编辑适配器；未被路由的非直接提示使用单独的局部性适配器，该适配器训练为保留或恢复原始对象偏好。我们在三个1,000案例协议\cf{}、\zsre{}和\mquake{}上，在相同记忆协议和两个7B/8B基础模型下评估\method{}。在Llama-3.1-8B-Instruct上，\method{}在所有三个基准上获得最佳总体概率偏好准确率：\cf{}为0.8180，\zsre{}为0.8946，\mquake{}为0.9922。在Qwen3-8B上趋势相同。路由器消融实验表明，相关记忆边界因数据集而异：在\cf{}上，词汇神经路由器最安全；而在\zsre{}和\mquake{}上，BGE嵌入路由效果更好。组件和模块消融实验表明，增益主要来自将编辑注入与离路抑制分离，而非单纯增加LoRA容量。
英文摘要：
Knowledge editing systems must update selected facts while preserving nearby but irrelevant behavior. This paper studies this problem in a memory-assisted setting where an edit memory is retrieved at inference time and a parameter-efficient adapter corrects the model's object preference. We argue that the central design question is not only how to write an edit, but also when to suppress it. We introduce RRDA, a route-specialized dual-adapter editor. A relevance router first decides whether a prompt should receive an edit memory. Routed prompts use an edit adapter trained to prefer the new object over the original object; unrouted non-direct prompts use a separate locality adapter trained to preserve or restore the original-object preference. We evaluate RRDA on three 1,000-case protocols, CounterFact, ZsRE, and MQuAKE-CF, under the same memory protocol and two 7B/8B base models. On Llama-3.1-8B-Instruct, RRDA obtains the best overall probability-preference accuracy on all three benchmarks: 0.8180 on CounterFact, 0.8946 on ZsRE, and 0.9922 on MQuAKE-CF. The same trend holds on Qwen3-8B. Router ablations show that the relevant memory boundary differs across datasets: a lexical neural router is safest on CounterFact, while BGE embedding routing is better on ZsRE and MQuAKE-CF. Memory, component, and module ablations show that explicit memory supplies the largest edit gain, while route-specialized adapters improve the final reliability-locality balance rather than simply increasing LoRA capacity.
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。
