# Uncovering and Mitigating Transient Blindness in Multimodal Model Editing

arXiv: https://arxiv.org/abs/2511.13243 · HTML: https://mlsanity.com/paper/2511.13243/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 22 of its 33 references that are in the map.
- First posted on arXiv 17 Nov 2025.

## Builds on (most cited first, 22 of 22)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](https://mlsanity.com/paper/1908.10084/) (2019)
- [Editing Factual Knowledge in Language Models](https://mlsanity.com/paper/2104.08164/) (2021)
- [Memory-Based Model Editing at Scale](https://mlsanity.com/paper/2206.06520/) (2022)
- [Multimodal Learning with Transformers: A Survey](https://mlsanity.com/paper/2206.06488/) (2022)
- [Can We Edit Factual Knowledge by In-Context Learning?](https://mlsanity.com/paper/2305.12740/) (2023)
- [Transformer-Patcher: One Mistake worth One Neuron](https://mlsanity.com/paper/2301.09785/) (2023)
- [Editable Neural Networks](https://mlsanity.com/paper/2004.00345/) (2020)
- [Strengthening Multimodal Large Language Model with Bootstrapped Preference Optimization](https://mlsanity.com/paper/2403.08730/) (2024)
- [Detecting Edit Failures In Large Language Models: An Improved Specificity Benchmark](https://mlsanity.com/paper/2305.17553/) (2023)
- [Learning to Edit: Aligning LLMs with Knowledge Editing](https://mlsanity.com/paper/2402.11905/) (2024)
- [Lifelong Knowledge Editing for LLMs with Retrieval-Augmented Continuous Prompt Learning](https://mlsanity.com/paper/2405.03279/) (2024)
- [VLKEB: A Large Vision-Language Model Knowledge Editing Benchmark](https://mlsanity.com/paper/2403.07350/) (2024)
- [MMKE-Bench: A Multimodal Editing Benchmark for Diverse Visual Knowledge](https://mlsanity.com/paper/2502.19870/) (2025)
- [Atomic Fact Decomposition Helps Attributed Question Answering](https://mlsanity.com/paper/2410.16708/) (2024)
- [Robust Disentangled Counterfactual Learning for Physical Audiovisual Commonsense Reasoning](https://mlsanity.com/paper/2502.12425/) (2025)
- [Prompting Large Language Models with Knowledge Graphs for Question Answering Involving Long-tail Facts](https://mlsanity.com/paper/2405.06524/) (2024)
- [Lifelong Knowledge Editing for Vision Language Models with Low-Rank Mixture-of-Experts](https://mlsanity.com/paper/2411.15432/) (2024)
- [ComprehendEdit: A Comprehensive Dataset and Evaluation Framework for Multimodal Knowledge Editing](https://mlsanity.com/paper/2412.12821/) (2024)
- [Long-Form Information Alignment Evaluation Beyond Atomic Facts](https://mlsanity.com/paper/2505.15792/) (2025)
- [Prompting Large Language Models with Partial Knowledge for Answering Questions with Unseen Entities](https://mlsanity.com/paper/2508.01290/) (2025)
- [Memorization neq Understanding: Do Large Language Models Have the Ability of Scenario Cognition?](https://mlsanity.com/paper/2509.04866/) (2025)

## Abstract (arXiv)

Multimodal Model Editing (MMED) aims to correct erroneous knowledge in multimodal models. Existing evaluation methods, adapted from textual model editing, overstate success by relying on low-similarity or random inputs, obscure overfitting. We propose a comprehensive locality evaluation framework, covering three key dimensions: random-image locality, no-image locality, and consistent-image locality, operationalized through seven distinct data types, enabling a detailed and structured analysis of multimodal edits. We introduce De-VQA, a dynamic evaluation for visual question answering, uncovering a phenomenon we term transient blindness, overfitting to edit-similar text while ignoring visuals. Token analysis shows edits disproportionately affect textual tokens. We propose locality-aware adversarial losses to balance cross-modal representations. Empirical results demonstrate that our approach consistently outperforms existing baselines, reducing transient blindness and improving locality by 17% on average.