More Edits, More Stable: Understanding the Lifelong Normalization in Sequential Model Editing
For researchers working on lifelong model editing of large language models, this work provides the first theoretical understanding of a critical but poorly understood component (LN) and offers a practical improvement.
The paper identifies Lifelong Normalization (LN) as a key mechanism for stability in sequential model editing of LLMs, provides a theoretical analysis showing it creates a self-reinforcing stability loop with asymptotic orthogonality and bounded norms, and proposes StableEdit which improves long-horizon stability via warm-up and whitening, achieving competitive performance.
Lifelong Model Editing aims to continuously update evolving facts in Large Language Models while preserving unrelated knowledge and general capabilities, yet it remains plagued by catastrophic forgetting and model collapse. Empirically, we find that recent editors resilient over long horizons share the same core strategy: Lifelong Normalization (LN), which normalizes value gradients using running statistics. Removing LN causes immediate performance collapse, and we observe a counter-intuitive positive cumulative effect where early edits can promote the success of future edits. Yet the mechanism of LN remains a "black box", leaving its precise role in lifelong stability poorly understood. In this work, we provide the first theoretical account of LN in the lifelong regime. Our analysis reveals a self-reinforcing stability loop and proves that, when combined with ridge-regularized regression, LN yields parameter updates with asymptotic orthogonality and bounded norms, directly mitigating forgetting and systemic collapse. Based on these insights, we derive StableEdit, which strengthens this stability loop via an explicit warm-up stage and full whitening, improving long-horizon stability at minimal overhead. Extensive experiments validate our theory and demonstrate competitive performance. Our code is available at https://github.com/MINE-USTC/StableEdit.