EVLM: An Efficient Vision-Language Model for Visual Understanding
This addresses efficiency and perception issues in multi-modal AI for tasks like image and video captioning, but it is incremental as it builds on existing architectures like LLaVA and Flamingo.
The paper tackles the computational overhead and limited visual perception in multi-modal language models by proposing an efficient vision-language model that uses cross-attention, hierarchical ViT features, and Mixture of Experts, achieving competitive scores on public benchmarks.
In the field of multi-modal language models, the majority of methods are built on an architecture similar to LLaVA. These models use a single-layer ViT feature as a visual prompt, directly feeding it into the language models alongside textual tokens. However, when dealing with long sequences of visual signals or inputs such as videos, the self-attention mechanism of language models can lead to significant computational overhead. Additionally, using single-layer ViT features makes it challenging for large language models to perceive visual signals fully. This paper proposes an efficient multi-modal language model to minimize computational costs while enabling the model to perceive visual signals as comprehensively as possible. Our method primarily includes: (1) employing cross-attention to image-text interaction similar to Flamingo. (2) utilize hierarchical ViT features. (3) introduce the Mixture of Experts (MoE) mechanism to enhance model effectiveness. Our model achieves competitive scores on public multi-modal benchmarks and performs well in tasks such as image captioning and video captioning.