FlashVLM: Text-Guided Visual Token Selection for Large Multimodal Models

TL;DR

FlashVLM introduces a text-guided visual token selection method that dynamically reduces visual input size in large multimodal models, achieving high compression with minimal accuracy loss and improving efficiency without sacrificing robustness.

Contribution

It proposes a novel cross-modal similarity-based token selection framework that outperforms existing methods in visual token reduction while maintaining semantic alignment and model performance.

Findings

Prunes up to 77.8% of visual tokens with minimal accuracy loss.

Achieves state-of-the-art efficiency-performance trade-offs.

Maintains robustness and generalization across multiple benchmarks.

Abstract

Large vision-language models (VLMs) typically process hundreds or thousands of visual tokens per image or video frame, incurring quadratic attention cost and substantial redundancy. Existing token reduction methods often ignore the textual query or rely on deep attention maps, whose instability under aggressive pruning leads to degraded semantic alignment. We propose FlashVLM, a text guided visual token selection framework that dynamically adapts visual inputs to the query. Instead of relying on noisy attention weights, FlashVLM computes an explicit cross modal similarity between projected image tokens and normalized text embeddings in the language model space. This extrinsic relevance is fused with intrinsic visual saliency using log domain weighting and temperature controlled sharpening. In addition, a diversity preserving partition retains a minimal yet representative set of background tokens to maintain global context. Under identical token budgets and evaluation protocols, FlashVLM achieves beyond lossless compression, slightly surpassing the unpruned baseline while pruning up to 77.8 percent of visual tokens on LLaVA 1.5, and maintaining 92.8 percent accuracy even under 94.4 percent compression. Extensive experiments on 14 image and video benchmarks demonstrate that FlashVLM delivers state of the art efficiency performance trade offs while maintaining strong robustness and generalization across mainstream VLMs.