VoCo-LLaMA: Towards Vision Compression with Large Language Models

TL;DR

VoCo-LLaMA introduces a novel vision token compression method using large language models, significantly reducing computational costs while maintaining performance, and effectively understanding temporal correlations in videos.

Contribution

It is the first approach to compress vision tokens with LLMs, leveraging attention distillation to improve efficiency and temporal understanding in multi-modal tasks.

Findings

Achieves 576× compression ratio with minimal performance loss.

Reduces inference FLOPs by up to 94.8%.

Outperforms previous methods on video question-answering benchmarks.

Abstract

Vision-Language Models (VLMs) have achieved remarkable success in various multi-modal tasks, but they are often bottlenecked by the limited context window and high computational cost of processing high-resolution image inputs and videos. Vision compression can alleviate this problem by reducing the vision token count. Previous approaches compress vision tokens with external modules and force LLMs to understand the compressed ones, leading to visual information loss. However, the LLMs' understanding paradigm of vision tokens is not fully utilised in the compression learning process. We propose VoCo-LLaMA, the first approach to compress vision tokens using LLMs. By introducing Vision Compression tokens during the vision instruction tuning phase and leveraging attention distillation, our method distill how LLMs comprehend vision tokens into their processing of VoCo tokens. VoCo-LLaMA facilitates effective vision compression and improves the computational efficiency during the inference stage. Specifically, our method achieves minimal performance loss with a compression ratio of 576$\times$, resulting in up to 94.8$\%$ fewer FLOPs and 69.6$\%$ acceleration in inference time. Furthermore, through continuous training using time-series compressed token sequences of video frames, VoCo-LLaMA demonstrates the ability to understand temporal correlations, outperforming previous methods on popular video question-answering benchmarks. Our approach presents a promising way to unlock the full potential of VLMs' contextual window, enabling more scalable multi-modal applications. The project page, along with the associated code, can be accessed via https://yxxxb.github.io/VoCo-LLaMA-page/.