ACL Findings2026
DKME: Rethinking Coupled Knowledge Memory for Lifelong Model Editing of Large Language Models
Guanyu Zheng, Zhenyu Wang, Tingting He, Xv Wang, Haochang Wang, Yaokai Huang, Tiejun Zhao
Abstract
Lifelong knowledge editing aims to inject a stream of factual updates into large language models (LLMs) without retraining, yet existing memory-based editors often suffer from catastrophic forgetting as edits accumulate. We argue that a key factor is the coupled knowledge memory mechanism, where addressing (routing) and storage (writing via memorymodule updates) are entangled. This entanglement makes it difficult to confine the effects of each edit to its intended scope, particularly in multi-domain and associated-fact editing streams, where updates either span diverse semantic domains or repeatedly modify related attributes of the same subject. Consequently, updating memory for one edit inadvertently alters the routing and stored representations of previously injected edits, leading to catastrophic forgetting as edits accumulate. We propose DKME, which decouples addressing from storage via two stages: decoupled semantic addressing learns a factaware manifold for scope-aware routing, and partitioned memory storage localizes edits to memory partitions identified by unsupervised clustering in the embedding space. Experiments on three benchmarks, including HalluEd-itBench, CKnowEdit, and WikiData counterfact , demonstrate that DKME consistently achieves a more favorable trade-off between editing success and locality compared to baselines, while maintaining more stable performance as the edit scale increases.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
Lineage
Builds on26
- A Simple Framework for Contrastive Learning of Visual RepresentationsTing Chen, Simon Kornblith, Mohammad Norouzi, Geoffrey E. HintonICML 2020 · 24,064 citations
- Chain-of-Thought Prompting Elicits Reasoning in Large Language ModelsJason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma et al.NeurIPS 2022 · 22,562 citations
- Locating and Editing Factual Associations in GPTKevin Meng, David Bau, Alex Andonian, Yonatan BelinkovNeurIPS 2022 · 3,415 citations
- Post-Training Quantization for Vision TransformerZhenhua Liu, Yunhe Wang, Kai Han, Wei Zhang et al.NeurIPS 2021 · 528 citations
- Memory-Based Model Editing at ScaleEric Mitchell, Charles Lin, Antoine Bosselut, Christopher D. Manning et al.ICML 2022 · 520 citations
Related papers
- Knowledge Decoupling via Orthogonal Projection for Lifelong Editing of Large Language ModelsHaoyu Xu, Pengxiang Lan, Enneng Yang, Guibing Guo et al.ACL 2025 · 4 citations
- Disentangling Knowledge Representations for Large Language Model EditingMengqi Zhang, Zisheng Zhou, Xiaotian Ye, Qiang Liu et al.ICLR 2026 · 6 citations
- WikiBigEdit: Understanding the Limits of Lifelong Knowledge Editing in LLMsLukas Thede, Karsten Roth, Matthias Bethge, Zeynep Akata et al.ICML 2025
- Fisher-Driven Adaptive Locating for Knowledge Editing in Large Language ModelsChenghao Xu, Jiexi Yan, Guangtao Lyu, Qi Liu et al.ACL 2026
- DSCA: Dynamic Subspace Concept Alignment for Lifelong VLM EditingGyanendra Das, Sai Satyam JenaCVPR 2026 · 2 citations
