Language Models Can See Better: Visual Contrastive Decoding For LLM Multimodal Reasoning
This addresses the problem of resource-intensive training for multimodal tasks, offering a method to enhance LLMs' visual perception incrementally.
The paper tackles the challenge of enabling Large Language Models (LLMs) to perform multimodal reasoning without additional training by proposing the Modular-based Visual Contrastive Decoding (MVCD) framework, which improves model accuracy across five question answering datasets.
Although Large Language Models (LLMs) excel in reasoning and generation for language tasks, they are not specifically designed for multimodal challenges. Training Multimodal Large Language Models (MLLMs), however, is resource-intensive and constrained by various training limitations. In this paper, we propose the Modular-based Visual Contrastive Decoding (MVCD) framework to move this obstacle. Our framework leverages LLMs' In-Context Learning (ICL) capability and the proposed visual contrastive-example decoding (CED), specifically tailored for this framework, without requiring any additional training. By converting visual signals into text and focusing on contrastive output distributions during decoding, we can highlight the new information introduced by contextual examples, explore their connections, and avoid over-reliance on prior encoded knowledge. MVCD enhances LLMs' visual perception to make it see and reason over the input visuals. To demonstrate MVCD's effectiveness, we conduct experiments with four LLMs across five question answering datasets. Our results not only show consistent improvement in model accuracy but well explain the effective components inside our decoding strategy. Our code will be available at https://github.com/Pbhgit/MVCD.