Skip to yearly menu bar Skip to main content


Poster

Dense Connector for MLLMs

Huanjin Yao · Wenhao Wu · Taojiannan Yang · YuXin Song · Mengxi Zhang · Haocheng Feng · Yifan Sun · Zhiheng Li · Wanli Ouyang · Jingdong Wang

[ ]
Wed 11 Dec 11 a.m. PST — 2 p.m. PST

Abstract:

Do we fully leverage the potential of visual encoder in Multimodal Large Language Models (MLLMs)? The recent outstanding performance of MLLMs in multimodal understanding has garnered broad attention from both academia and industry. In the current MLLM rat race, the focus seems to be predominantly on the linguistic side. We witness the rise of larger and higher-quality instruction datasets, as well as the involvement of larger-sized LLMs. Yet, scant attention has been directed towards the visual signals utilized by MLLMs, often assumed to be the final high-level features extracted by a frozen visual encoder. In this paper, we introduce the Dense Connector - a simple, effective, and plug-and-play vision-language connector that significantly enhances existing MLLMs by leveraging multi-layer visual features, with minimal additional computational overhead. Furthermore, our model, trained solely on images, showcases remarkable zero-shot capabilities in video understanding as well. Experimental results across various vision encoders, image resolutions, training dataset scales, varying sizes of LLMs (2.7B→70B), and diverse architectures of MLLMs (e.g., LLaVA and Mini-Gemini) validate the versatility and scalability of our approach, achieving state-of-the-art performance on across 19 image and video benchmarks. We hope that this work will provide valuable experience and serve as a basic module for future MLLM development. Code is available at https://anonymous.4open.science/r/DC-NIPS.

Live content is unavailable. Log in and register to view live content