mlsanityData to 14 Oct 2026

Towards Localized and Disentangled Knowledge Editing for Multimodal Large Language Models

arXiv 2605.29826 ↗First posted 28 May 2026cs.CL cs.AIcited by 0

On the map: Multimodal LLMs / vision-language models → Image editing and quality assessment

Abstract (arXiv)

Existing methods in Multimodal Knowledge Editing (MKE) have advanced the ability to correct outdated or inaccurate knowledge in Multimodal Large Language Models (MLLMs). However, they exhibit a critical limitation: while effectively modifying target factual pairs, they fail to generalize edits to logically related queries and often cause unintended alterations to unrelated but visually or semantically linked information. We identify and formalize two underlying failure modes causing this issue: Causal Misalignment, which confines edits to the specific sample, and Feature Entanglement, which causes unintended alterations to coupled but irrelevant information. To address these issues, we propose Localized and Disentangled Knowledge Editing (LDKE), a new framework that achieves precise and generalized editing by localizing fact-specific model layers and disentangling target-relevant inputs from irrelevant ones. Our approach introduces a Fast Localization module to identify and update critical layers efficiently, along with a Disentanglement Classifier that routes inputs appropriately to preserve unrelated knowledge. Extensive experiments across various benchmarks and MLLMs demonstrate that LDKE achieves superior performance in propagating edits to related contexts while maintaining high locality.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Image editing and quality assessment, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 11 of its 27 references that are in the map. First posted on arXiv 28 May 2026.

Builds on

11 of 27 references.

References · 11

  1. 2025Gemma 3 Technical Reportcited by 1,824
  2. 2025InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiencycited by 1,510
  3. 2022Mass-Editing Memory in a Transformercited by 1,001
  4. 2023MM-REACT: Prompting ChatGPT for Multimodal Reasoning and Actioncited by 517
  5. 2021Fast Model Editing at Scalecited by 498
  6. 2020Modifying Memories in Transformer Modelscited by 224
  7. 2023Transformer-Patcher: One Mistake worth One Neuroncited by 197
  8. 2024A Comprehensive Study of Knowledge Editing for Large Language Modelscited by 163
  9. 2024Fine-grained Hallucination Detection and Editing for Language Modelscited by 149
  10. 2023Knowledge Sanitization of Large Language Modelscited by 37
  11. 2025Causal Tracing of Object Representations in Large Vision Language Models: Mechanistic Interpretability and Hallucination Mitigationcited by 23

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2024MC-MKE: A Fine-Grained Multimodal Knowledge Editing Benchmark Emphasizing Modality Consistency
  2. 2024ComprehendEdit: A Comprehensive Dataset and Evaluation Framework for Multimodal Knowledge Editing
  3. 2024Visual-Oriented Fine-Grained Knowledge Editing for MultiModal Large Language Models
  4. 2024Uncovering Overfitting in Large Language Model Editing
  5. 2026Generalizable Multimodal Large Language Model Editing via Invariant Trajectory Learning
  6. 2025Benchmarking and Rethinking Knowledge Editing for Large Language Models
  7. 2025Aligning Language Models with Real-time Knowledge Editing
  8. 2026Beyond Binary Edits Robust Multimodal Knowledge Editing with Adversarial Subspace Alignment