Revealing and Mitigating Over-Attention in Knowledge Editing

• URL: http://arxiv.org/abs/2502.14838v1
• Date: Thu, 20 Feb 2025 18:51:12 GMT
• Title: Revealing and Mitigating Over-Attention in Knowledge Editing
• Authors: Pinzheng Wang, Zecheng Tang, Keyan Zhou, Juntao Li, Qiaoming Zhu, Min Zhang,
• Abstract summary: Large Language Models have demonstrated superior performance across a wide range of tasks.<n>However, they still exhibit undesirable errors due to incorrect knowledge learned from the training data.<n> knowledge editing methods emerged to precisely edit the specific model knowledge via efficiently modifying a very small percentage of parameters.<n>These editing methods can lead to the problem of Specificity Failure, where the existing knowledge and capabilities are severely degraded due to editing.
• Score: 28.950187006528783
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large Language Models have demonstrated superior performance across a wide range of tasks, but they still exhibit undesirable errors due to incorrect knowledge learned from the training data. To avoid this, knowledge editing methods emerged to precisely edit the specific model knowledge via efficiently modifying a very small percentage of parameters. % However, those methods can lead to the problem of Specificity Failure: when the content related to the edited knowledge occurs in the context, it can inadvertently corrupt other pre-existing knowledge. However, those methods can lead to the problem of Specificity Failure, where the existing knowledge and capabilities are severely degraded due to editing. Our preliminary indicates that Specificity Failure primarily stems from the model's attention heads assigning excessive attention scores to entities related to the edited knowledge, thereby unduly focusing on specific snippets within the context, which we denote as the Attention Drift phenomenon. To mitigate such Attention Drift issue, we introduce a simple yet effective method Selective Attention Drift Restriction}(SADR), which introduces an additional regularization term during the knowledge editing process to restrict changes in the attention weight distribution, thereby preventing undue focus on the edited entity. Experiments on five frequently used strong LLMs demonstrate the effectiveness of our method, where SADR can significantly mitigate Specificity Failure in the predominant knowledge editing tasks.

Related papers

• Are We Evaluating the Edit Locality of LLM Model Editing Properly? [68.441768731381]
  We find that existing specificity evaluation protocols are inadequate for this purpose.<n>Existing specificity metrics are weakly correlated with the strength of specificity regularizers.<n>We also find that current metrics lack sufficient sensitivity, rendering them ineffective at distinguishing the specificity performance of different methods.
  arXiv  Detail & Related papers  (2026-01-24T07:07:21Z)
• Retention analysis of edited knowledge after fine-tuning [5.440397659472036]
  Large language models (LLMs) store vast amounts of knowledge, which often requires updates to correct factual errors, incorporate newly acquired information, or adapt model behavior.<n>Model editing methods have emerged as efficient solutions for such updates, offering localized and precise knowledge modification at significantly lower computational cost than continual training.<n>However, the effect of fine-tuning on previously edited knowledge remains poorly understood.
  arXiv  Detail & Related papers  (2025-07-14T15:51:19Z)
• ThinkEval: Practical Evaluation of Knowledge Leakage in LLM Editing using Thought-based Knowledge Graphs [3.9295613363026174]
  We present ThinkEval, a framework to quantify indirect knowledge leakage and ripple effects in model-editing.<n>ThinkEval builds and employs specialized knowledge graphs to analyze the causal structure of facts before and after editing.<n>We evaluate five editing techniques: AlphaEdit, RECT, ROME, MEMIT, and PRUNE.
  arXiv  Detail & Related papers  (2025-06-02T07:24:12Z)
• Precise Localization of Memories: A Fine-grained Neuron-level Knowledge Editing Technique for LLMs [47.06544781855325]
  We propose a Fine-grained Neuron-level Knowledge Editing (FiNE) method that enhances editing locality without affecting success rates. By precisely identifying and modifying specific neurons within feed-forward networks, FiNE significantly improves knowledge localization and editing.
  arXiv  Detail & Related papers  (2025-03-03T01:30:28Z)
• Related Knowledge Perturbation Matters: Rethinking Multiple Pieces of Knowledge Editing in Same-Subject [49.559994791305535]
  Current state-of-the-art editing methods struggle when tasked with editing multiple related knowledge pieces for the same subject.<n>We introduce the $textS2textRKE$(Same-Subject Related Knowledge Editing) benchmark.<n>Our experiments reveal that only mainstream locate-then-edit methods, such as ROME and MEMIT, exhibit "related knowledge perturbation"
  arXiv  Detail & Related papers  (2025-02-08T04:47:17Z)
• Uncovering Overfitting in Large Language Model Editing [35.55260822503773]
  We identify and investigate the phenomenon of Editing Overfit, where edited models assign disproportionately high probabilities to the edit target. We propose a new plug-and-play strategy called Learn to Inference (LTI), which introduce a Multi-stage Inference Constraint module to guide the edited models in recalling new knowledge.
  arXiv  Detail & Related papers  (2024-10-10T11:09:00Z)
• Mind the Interference: Retaining Pre-trained Knowledge in Parameter Efficient Continual Learning of Vision-Language Models [79.28821338925947]
  Domain-Class Incremental Learning is a realistic but challenging continual learning scenario. To handle these diverse tasks, pre-trained Vision-Language Models (VLMs) are introduced for their strong generalizability. This incurs a new problem: the knowledge encoded in the pre-trained VLMs may be disturbed when adapting to new tasks, compromising their inherent zero-shot ability. Existing methods tackle it by tuning VLMs with knowledge distillation on extra datasets, which demands heavy overhead. We propose the Distribution-aware Interference-free Knowledge Integration (DIKI) framework, retaining pre-trained knowledge of
  arXiv  Detail & Related papers  (2024-07-07T12:19:37Z)
• Outdated Issue Aware Decoding for Reasoning Questions on Edited Knowledge [93.54427119091174]
  We propose outDated ISsue aware deCOding to enhance the performance of edited models on reasoning questions. We capture the difference in the probability distribution between the original and edited models. We amplify the difference of the token prediction in the edited model to alleviate the outdated issue.
  arXiv  Detail & Related papers  (2024-06-05T03:00:15Z)
• EVEDIT: Event-based Knowledge Editing with Deductive Editing Boundaries [69.72012539060731]
  We introduce a theoretical framework for efficient knowledge editing (KE) in large language models (LLMs) We propose a novel task of event-based knowledge editing that pairs facts with event descriptions. We empirically demonstrate the superiority of event-based editing over the existing setting on resolving uncertainty in edited models.
  arXiv  Detail & Related papers  (2024-02-17T16:34:50Z)
• A Comprehensive Study of Knowledge Editing for Large Language Models [82.65729336401027]
  Large Language Models (LLMs) have shown extraordinary capabilities in understanding and generating text that closely mirrors human communication. This paper defines the knowledge editing problem and provides a comprehensive review of cutting-edge approaches. We introduce a new benchmark, KnowEdit, for a comprehensive empirical evaluation of representative knowledge editing approaches.
  arXiv  Detail & Related papers  (2024-01-02T16:54:58Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.