发表机构
Shanghai AI Laboratory; Tsinghua University（上海人工智能实验室; 清华大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
研究针对大语言模型应用于特定领域产生的对齐代价问题，提出MemSFT方法，通过参数内存解耦领域专业化与主干参数更新，经多领域多模型评估，能提升领域性能且通用性能下降小，实现通用与专业能力解耦。
AI 中文摘要
将大语言模型（LLMs）应用于特定领域通常会产生对齐代价，因为在特定领域任务上微调会导致灾难性遗忘并大幅降低通用任务性能。我们提出了MemSFT，它通过即插即用的参数内存将领域专业化与主干参数更新解耦，从而减轻对齐代价。该内存经过训练以模仿对领域数据运行的非参数检索器的行为，从而记住原本通过检索获取的知识和模式。一旦在特定领域训练，该内存可在不同大小的LLMs间复用。生成时，学习到的路由器在每个解码步骤动态融合内存和主干的输出分布，有选择地调用领域专业知识。在生物学、地球科学和法律领域，对从Qwen3 - 8B到Qwen3 - 235B - A22B等模型的评估表明，MemSFT持续提升领域性能，通用性能下降可忽略不计，而完全微调在通用任务上会严重遗忘。总体而言，我们的结果展示了在参数层面将通用模型能力与特定领域知识解耦的实用途径，使LLMs在不损害通用能力的情况下具备新的专业能力。
英文摘要
Adapting Large Language Models (LLMs) to specialized domains often incurs an alignment tax, as fine-tuning on domain-specific tasks can cause catastrophic forgetting and substantially degrade performance on general tasks. We propose MemSFT, which mitigates the alignment tax by decoupling domain specialization from backbone parameter updates through a plug-and-play parametric memory. The memory is trained to imitate the behavior of a non-parametric retriever operating over domain data, thereby memorizing knowledge and patterns that would otherwise be accessed through retrieval. Once trained on a specific domain, the memory can be reused across LLMs of different sizes. During generation, a learned router dynamically fuses the output distributions of the memory and backbone at each decoding step, allowing domain expertise to be invoked selectively. Across biology, geoscience, and law, evaluations with models ranging from Qwen3-8B to Qwen3-235B-A22B show that MemSFT consistently improves domain performance with negligible degradation in general performance, whereas full SFT suffers severe forgetting on general tasks. Overall, our results demonstrate a practical path to decoupling general model capabilities from domain-specific knowledge at the parameter level, thereby equipping LLMs with new specialized capabilities without compromising their general capabilities.
Comments33 pages, 11 figures, 13 tables
