Attribution Analysis Meets Model Editing: Advancing Knowledge Correction in Vision Language Models with VisEdit

• URL: http://arxiv.org/abs/2408.09916v3
• Date: Thu, 23 Jan 2025 11:03:13 GMT
• Title: Attribution Analysis Meets Model Editing: Advancing Knowledge Correction in Vision Language Models with VisEdit
• Authors: Qizhou Chen, Taolin Zhang, Chengyu Wang, Xiaofeng He, Dakan Wang, Tingting Liu,
• Abstract summary: We use contribution allocation and noise perturbation methods to measure the contributions of visual representations for token predictions.<n>Our attribution analysis shows that visual representations in mid-to-later layers that are highly relevant to the prompt contribute significantly to predictions.<n>We propose VisEdit, a novel model editor for Vision-LLMs that effectively corrects knowledge by editing intermediate visual representations in regions important to the edit prompt.
• Score: 18.71195974474024
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Model editing aims to correct outdated or erroneous knowledge in large models without costly retraining. Recent research discovered that the mid-layer representation of the subject's final token in a prompt has a strong influence on factual predictions, and developed Large Language Model (LLM) editing techniques based on this observation. However, for Vision-LLMs (VLLMs), how visual representations impact the predictions from a decoder-only language model remains largely unexplored. To the best of our knowledge, model editing for VLLMs has not been extensively studied in the literature. In this work, we employ the contribution allocation and noise perturbation methods to measure the contributions of visual representations for token predictions. Our attribution analysis shows that visual representations in mid-to-later layers that are highly relevant to the prompt contribute significantly to predictions. Based on these insights, we propose VisEdit, a novel model editor for VLLMs that effectively corrects knowledge by editing intermediate visual representations in regions important to the edit prompt. We evaluated VisEdit using multiple VLLM backbones and public VLLM editing benchmark datasets. The results show the superiority of VisEdit over the strong baselines adapted from existing state-of-the-art editors for LLMs.

Related papers

• Latent Knowledge Scalpel: Precise and Massive Knowledge Editing for Large Language Models [3.834827405473377]
  Large Language Models (LLMs) often retain inaccurate or outdated information from pre-training, leading to incorrect predictions or biased outputs during inference.<n>We introduce the Latent Knowledge Scalpel (LKS), an LLM editor that manipulates the latent knowledge of specific entities via a lightweight hypernetwork to enable precise and large-scale editing.<n> Experiments conducted on Llama-2 and Mistral show even with the number of simultaneous edits reaching 10,000, LKS effectively performs knowledge editing while preserving the general abilities of the edited LLMs.
  arXiv  Detail & Related papers  (2025-08-01T03:51:43Z)
• DualEdit: Dual Editing for Knowledge Updating in Vision-Language Models [26.762431651154607]
  We propose DualEdit, an editor that modifies both textual and visual modalities at their respective key layers.<n>We evaluate DualEdit across multiple VLM backbones and benchmark datasets, demonstrating its superiority over state-of-the-art VLM editing baselines.
  arXiv  Detail & Related papers  (2025-06-16T16:04:16Z)
• InComeS: Integrating Compression and Selection Mechanisms into LLMs for Efficient Model Editing [77.47790551485721]
  In-context learning is a promising editing method by comprehending edit information through context encoding.<n>This method is constrained by the limited context window of large language models.<n>We propose InComeS, a flexible framework that enhances LLMs' ability to process editing contexts.
  arXiv  Detail & Related papers  (2025-05-28T09:20:18Z)
• Envisioning Beyond the Pixels: Benchmarking Reasoning-Informed Visual Editing [84.16442052968615]
  We introduce RISEBench, the first benchmark for evaluating Reasoning-Informed viSual Editing (RISE)<n>RISEBench focuses on four key reasoning categories: Temporal, Causal, Spatial, and Logical Reasoning.<n>We conduct experiments evaluating nine prominent visual editing models, comprising both open-source and proprietary models.
  arXiv  Detail & Related papers  (2025-04-03T17:59:56Z)
• Lifelong Knowledge Editing for Vision Language Models with Low-Rank Mixture-of-Experts [17.376346967267327]
  We propose LiveEdit, a LIfelong Vision language modEl Edit to bridge the gap between lifelong LLM editing and Vision LLM editing. A hard filtering mechanism is developed to utilize visual semantic knowledge, thereby eliminating visually irrelevant experts for input queries. To integrate visually relevant experts, we introduce a soft routing mechanism based on textual semantic relevance to achieve multi-expert fusion.
  arXiv  Detail & Related papers  (2024-11-23T03:19:40Z)
• Fine-Grained Verifiers: Preference Modeling as Next-token Prediction in Vision-Language Alignment [57.0121616203175]
  We propose FiSAO, a novel self-alignment method that utilizes the model's own visual encoder as a fine-grained verifier to improve vision-language alignment. By leveraging token-level feedback from the vision encoder, FiSAO significantly improves vision-language alignment, even surpassing traditional preference tuning methods that require additional data.
  arXiv  Detail & Related papers  (2024-10-18T03:34:32Z)
• VLKEB: A Large Vision-Language Model Knowledge Editing Benchmark [53.091690659399234]
  knowledge editing on large language models (LLMs) has received considerable attention. The existing LVLM editing benchmark, which comprises three metrics (Reliability, Locality, and Generality), falls short in the quality of synthesized evaluation images. We employ more reliable data collection methods to construct a new Large $textbfV$ision-$textbfL$anguage Model.
  arXiv  Detail & Related papers  (2024-03-12T06:16:33Z)
• Editing Conceptual Knowledge for Large Language Models [65.38231526537476]
  This paper pioneers the investigation of editing conceptual knowledge for Large Language Models (LLMs) We construct a novel benchmark dataset ConceptEdit and establish a suite of new metrics for evaluation. experimental results reveal that, although existing editing methods can efficiently modify concept-level definition to some extent, they also have the potential to distort the related instantial knowledge.
  arXiv  Detail & Related papers  (2024-03-10T16:57:10Z)
• Knowledge Graph Enhanced Large Language Model Editing [37.6721061644483]
  Large language models (LLMs) are pivotal in advancing natural language processing (NLP) tasks. Existing editing methods struggle to track and incorporate changes in knowledge associated with edits. We propose a novel model editing method that leverages knowledge graphs for enhancing LLM editing, namely GLAME.
  arXiv  Detail & Related papers  (2024-02-21T07:52:26Z)
• The Butterfly Effect of Model Editing: Few Edits Can Trigger Large Language Models Collapse [58.0132400208411]
  Even a single edit can trigger model collapse, manifesting as significant performance degradation in various benchmark tasks. benchmarking Large Language Models after each edit is impractically time-consuming and resource-intensive. We have utilized GPT-3.5 to develop a new dataset, HardEdit, based on hard cases.
  arXiv  Detail & Related papers  (2024-02-15T01:50:38Z)
• Model Editing Harms General Abilities of Large Language Models: Regularization to the Rescue [122.20016030723043]
  We evaluate the side effects of model editing on large language models (LLMs) Our analysis reveals that the side effects are caused by model editing altering the original model weights excessively. To mitigate this, a method named RECT is proposed to regularize the edit update weights.
  arXiv  Detail & Related papers  (2024-01-09T18:03:15Z)
• Memory-Based Model Editing at Scale [102.28475739907498]
  Existing model editors struggle to accurately model an edit's intended scope. We propose Semi-Parametric Editing with a Retrieval-Augmented Counterfactual Model (SERAC) SERAC stores edits in an explicit memory and learns to reason over them to modulate the base model's predictions as needed.
  arXiv  Detail & Related papers  (2022-06-13T23:40:34Z)

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.