Script: Graph-Structured and Query-Conditioned Semantic Token Pruning for Multimodal Large Language Models

TL;DR

Script is a versatile, plug-and-play token pruning method for multimodal large language models that improves efficiency and accuracy without retraining, by removing redundant and query-irrelevant visual tokens.

Contribution

It introduces a novel graph-structured and query-conditioned semantic pruning approach that enhances model efficiency and generalizes across diverse MLLMs without retraining.

Findings

Achieves up to 6.8x prefill speedup on LLaVA-NeXT-7B.

Reduces FLOPs by 10x while retaining 96.88% performance.

Outperforms existing pruning methods on 14 benchmarks.

Abstract

The rapid growth of visual tokens in multimodal large language models (MLLMs) leads to excessive memory consumption and inference latency, especially when handling high-resolution images and videos. Token pruning is a technique used to mitigate this issue by removing redundancy, but existing methods often ignore relevance to the user query or suffer from the limitations of attention mechanisms, reducing their adaptability and effectiveness. To address these challenges, we propose Script, a plug-and-play pruning method that requires no retraining and generalizes across diverse MLLMs. Script comprises two modules: a graph-structured pruning module that removes visually redundant tokens, and a query-conditioned semantic pruning module that preserves query-relevant visual information. Together, they enhance performance on multimodal tasks. Experiments on fourteen benchmarks across image and video understanding tasks show that Script consistently achieves higher model efficiency and predictive accuracy compared to existing pruning methods. On LLaVA-NeXT-7B, it achieves up to 6.8x prefill speedup and 10x FLOP reduction, while retaining 96.88% of the original performance.