English

Fwd2Bot: LVLM Visual Token Compression with Double Forward Bottleneck

Computer Vision and Pattern Recognition 2025-03-28 v1 Artificial Intelligence Machine Learning

Abstract

In this work, we aim to compress the vision tokens of a Large Vision Language Model (LVLM) into a representation that is simultaneously suitable for (a) generative and (b) discriminative tasks, (c) is nearly lossless, and (d) is storage-efficient. We propose a novel compression approach, called Fwd2Bot, that uses the LVLM itself to compress the visual information in a task-agnostic manner. At the core of Fwd2bot there exists a "double-forward pass" training strategy, whereby, during the first forward pass, the LLM (of the LVLM) creates a bottleneck by condensing the visual information into a small number of summary tokens. Then, using the same LLM, the second forward pass processes the language instruction(s) alongside the summary tokens, used as a direct replacement for the image ones. The training signal is provided by two losses: an autoregressive one applied after the second pass that provides a direct optimization objective for compression, and a contrastive loss, applied after the first pass, that further boosts the representation strength, especially for discriminative tasks. The training is further enhanced by stage-specific adapters. We accompany the proposed method by an in-depth ablation study. Overall, Fwd2Bot results in highly-informative compressed representations suitable for both generative and discriminative tasks. For generative tasks, we offer a 2x higher compression rate without compromising the generative capabilities, setting a new state-of-the-art result. For discriminative tasks, we set a new state-of-the-art on image retrieval and compositionality.

Keywords

Cite

@article{arxiv.2503.21757,
  title  = {Fwd2Bot: LVLM Visual Token Compression with Double Forward Bottleneck},
  author = {Adrian Bulat and Yassine Ouali and Georgios Tzimiropoulos},
  journal= {arXiv preprint arXiv:2503.21757},
  year   = {2025}
}
R2 v1 2026-06-28T22:37:04.516Z