提出一种可扩展的LLM编辑方法，有效避免修改时破坏模型通用能力。
CrispEdit: Low-Curvature Projections for Scalable Non-Destructive LLM Editing
- 通过低曲率子空间投影约束能力保持，实现精准编辑。
- 在多个基准上编辑成功率高，平均能力退化低于1%。
- 适合需要安全编辑且不牺牲模型泛化能力的研究者使用。
大语言模型编辑的核心挑战是能力保持：现有方法虽能改变特定行为，却可能通过操纵编辑代理导致通用能力退化，产生类似代理/奖励劫持的异常行为。本文提出CrispEdit，一种可扩展且原理严谨的二阶编辑算法，将能力保持作为显式约束，统一并推广了多种现有编辑方法。CrispEdit将编辑建模为带约束的优化问题，通过将更新投影到能力损失曲面的低曲率子空间来强制约束。其核心在于用Bregman散度表达能力约束，其二次形式恰好得到Gauss-Newton Hessian，即使基础模型未完全收敛也成立。我们利用Kronecker分解近似曲率（K-FAC）和一种新颖的无矩阵投影器，高效处理大规模模型。在标准模型编辑基准上，CrispEdit实现了高编辑成功率，同时在各数据集上平均能力退化低于1%，显著优于先前编辑器。
原文摘要 · Abstract (English)
A central challenge in large language model (LLM) editing is capability preservation: methods that successfully change targeted behavior can quietly game the editing proxy and corrupt general capabilities, producing degenerate behaviors reminiscent of proxy/reward hacking. We present CrispEdit, a scalable and principled second-order editing algorithm that treats capability preservation as an explicit constraint, unifying and generalizing several existing editing approaches. CrispEdit formulates editing as constrained optimization and enforces the constraint by projecting edit updates onto the low-curvature subspace of the capability-loss landscape. At the crux of CrispEdit is expressing capability constraint via Bregman divergence, whose quadratic form yields the Gauss-Newton Hessian exactly and even when the base model is not trained to convergence. We make this second-order procedure efficient at the LLM scale using Kronecker-factored approximate curvature (K-FAC) and a novel matrix-free projector that exploits Kronecker structure to avoid constructing massive projection matrices. Across standard model-editing benchmarks, CrispEdit achieves high edit success while keeping capability degradation below 1% on average across datasets, significantly improving over prior editors.
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。