English

Leveraging Visual Tokens for Extended Text Contexts in Multi-Modal Learning

Computer Vision and Pattern Recognition 2024-06-05 v1

Abstract

Training models with longer in-context lengths is a significant challenge for multimodal model due to substantial GPU memory and computational costs. This exploratory study does not present state-of-the-art models; rather, it introduces an innovative method designed to increase in-context text length in multi-modality large language models (MLLMs) efficiently. We present Visualized In-Context Text Processing (VisInContext), which processes long in-context text using visual tokens. This technique significantly reduces GPU memory usage and floating point operations (FLOPs) for both training and inferenceing stage. For instance, our method expands the pre-training in-context text length from 256 to 2048 tokens with nearly same FLOPs for a 56 billion parameter MOE model. Experimental results demonstrate that model trained with VisInContext delivers superior performance on common downstream benchmarks for in-context few-shot evaluation. Additionally, VisInContext is complementary to existing methods for increasing in-context text length and enhances document understanding capabilities, showing great potential in document QA tasks and sequential document retrieval.

Keywords

Cite

@article{arxiv.2406.02547,
  title  = {Leveraging Visual Tokens for Extended Text Contexts in Multi-Modal Learning},
  author = {Alex Jinpeng Wang and Linjie Li and Yiqi Lin and Min Li and Lijuan Wang and Mike Zheng Shou},
  journal= {arXiv preprint arXiv:2406.02547},
  year   = {2024}
}

Comments

12 pages. The website is \url{https://fingerrec.github.io/visincontext}

R2 v1 2026-06-28T16:53:20.040Z