Spectral Characterization and Mitigation of Sequential Knowledge Editing Collapse (2026.acl-long)
Copied to clipboard
Chi Zhang, Mengqi Zhang, Xiaotian Ye, Runxi Cheng, Zisheng Zhou, Ying Zhou, Pengjie Ren, Zhumin Chen
| Challenge: | Existing approaches to reducing the effects of knowledge editing are insufficiently understood. |
| Approach: | They propose a plug-and-play framework that preserves the dominant subspace of the original weights and analyzes parameter updates in the spectral basis of the weights. |
| Outcome: | The proposed framework improves editing efficacy while preserving general abilities under long-horizon sequential editing, including extreme settings with up to 20,000 edits. |
Similar Papers
Can We Continually Edit Language Models? On the Knowledge Attenuation in Sequential Model Editing (2024.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for model editing suffer from knowledge attenuation due to redundant parameters interference and update weight disentanglement. |
| Approach: | They propose a method to mitigate the problem of knowledge attenuation in sequential editing by analyzing redundant parameters interference and update weight disentanglement. |
| Outcome: | The proposed method mitigates the knowledge attenuation issue and improves on existing benchmarks. |
Lifelong Knowledge Editing requires Better Regularization (2025.findings-emnlp)
Copied to clipboard
Akshat Gupta, Phudish Prateepamornkul, Maochuan Lu, Ahmed Alaa, Thomas Hartvigsen, Gopala Anumanchipalli
| Challenge: | Knowledge editing is a promising way to improve factuality in large language models, but recent studies have shown significant model degradation during sequential editing. |
| Approach: | They formalize locate-then-edit methods as a two-step fine-tuning process . they show that model degradation occurs due to over-optimization of internal activations . |
| Outcome: | The proposed methods reduce time and improve factuality by 42-61%. |
LyapLock: Bounded Knowledge Preservation in Sequential Large Language Model Editing (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing models for enhancing knowledge updating are prone to performance degradation due to incomplete knowledge preservation mechanisms. |
| Approach: | They propose a model for locate-then-edit that decomposes long-term constrained programming into tractable stepwise subproblems for efficient solving. |
| Outcome: | The proposed framework achieves asymptotic optimal editing performance while meeting the constraints of long-term knowledge preservation. |
SpecEdit: A Spectral Approach for Multi-Round Knowledge Editing (2026.findings-acl)
Copied to clipboard
| Challenge: | Multi-round knowledge editing suffers from performance degradation as edits accumulate . intrinsic knowledge of model and historical edit memories are naively coupled during editing . SpecEdit improves model editing performance by reducing destructive coupling . |
| Approach: | They propose a spectral-based model editing module that integrates into existing editing methods without altering their original optimization procedures. |
| Outcome: | The proposed model improves performance on multiple LLMs and editing methods. |
Constraining Sequential Model Editing with Editing Anchor Compression (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large language models (LLMs) exhibit hallucinations due to incorrect or outdated knowledge embedded in their parameters. |
| Approach: | They propose a framework to constrain the deviation of the parameter matrix during sequential editing by selecting editing anchors that are important in encoding new relations without deviating too much from the original matrix. |
| Outcome: | The proposed framework minimizes deviations caused by model editing while retaining over 70% of the general abilities. |
Navigating the Dual Facets: A Comprehensive Evaluation of Sequential Memory Editing in Large Language Models (2024.acl-long)
Copied to clipboard
Zihao Lin, Mohammad Beigi, Hongxuan Li, Yufan Zhou, Yuxiang Zhang, Qifan Wang, Wenpeng Yin, Lifu Huang
| Challenge: | Memory Editing (ME) has emerged as an efficient method to modify erroneous facts or inject new knowledge into Large Language Models (LLMs). |
| Approach: | They propose to evaluate LLMs with single edit only and parameter-modifying ME with parameter-preserving ME. |
| Outcome: | The proposed method can maintain LLMs’ fundamental capabilities but struggles to accurately recall edited knowledge presented in a different format. |
The Butterfly Effect of Model Editing: Few Edits Can Trigger Large Language Models Collapse (2024.findings-acl)
Copied to clipboard
| Challenge: | Even a single edit can trigger model collapse, manifesting as significant performance degradation in various benchmark tasks. |
| Approach: | They propose to use perplexity as a surrogate metric to determine whether an edited model's performance is affected by a single edit. |
| Outcome: | The proposed method shows that even a single edit can cause model collapse, manifesting as significant performance degradation in various benchmark tasks. |
CaPEdit: Capability-Preserving Lifelong Knowledge Editing For Language Models (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to incrementally correct factual inaccuracies in large language models (LLMs) but sequential edits can lead to substantial degradation of capabilities. |
| Approach: | They propose a framework that preserves model capabilities under LKE by decoupling fast-updating factual knowledge from slow-evolving procedural knowledge. |
| Outcome: | The proposed framework improves capability preservation across all fundamental capabilities by 49.78% and achieves superior editing performance. |
Rebuilding ROME : Resolving Model Collapse during Sequential Model Editing (2024.emnlp-main)
Copied to clipboard
| Challenge: | Recent work using Rank-One Model Editing (ROME) has shown that there are certain facts that the algorithm is unable to edit without breaking the model. |
| Approach: | They propose to use a model editing method called Rank-One Model Editing to make multiple edits to a single model without breaking it. |
| Outcome: | The proposed method improves generalization and locality of model editing and improves model collapse compared to the original implementation of ROME. |
The Fall of ROME: Understanding the Collapse of LLMs in Model Editing (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Recent studies have found that model editing methods can cause large language models to collapse with just a single edit. |
| Approach: | They propose a method that uses prefixed keys and adds prefixes during testing to prevent model collapse. |
| Outcome: | The proposed method prevents model collapse while maintaining effectiveness, the authors show . Rank-One Model Editing (ROME) has been found to cause model collapse with just a single edit . |