Uncovering and Mitigating Transient Blindness in Multimodal Model Editing
This work addresses a critical evaluation gap in multimodal model editing, which is important for researchers and practitioners in AI, though it is incremental as it builds on existing methods.
The paper tackled the problem of overfitting in multimodal model editing by introducing a comprehensive evaluation framework and uncovering transient blindness, where models ignore visuals after edits. Their proposed locality-aware adversarial losses reduced transient blindness and improved locality by 17% on average.
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.