AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

VFlowOpt: A Token Pruning Framework for LMMs with Visual Information Flow-Guided Optimization

Sihan Yang, Runsen Xu, Chenhang Cui, Tai Wang, Dahua Lin, Jiangmiao Pang

ICCV 2025 · Vision-language and multimodal › Multimodal LLMs · approach: Attention-based visual token pruning · 2 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use VFlowOpt instead of VisionZip at 10% retentionLLaVA-OneVision-7B · 10 image benchmarks avgrelative avg performance (%): 82.1% -> 85.5%Improved
use VFlowOpt at 50% retentionLLaVA-OneVision-7B · 10 image benchmarks avgrelative avg performance (%): 98.9% (VisionZip) -> 99.9%Improved
add importance calibration (attention from relatively important tokens plus entropy) (ablation)VFlowOpt · MMStar (retain 25%)accuracy: 56.2 -> 57.8Improved
add token recycling (ablation)VFlowOpt · MMStar (retain 25%)accuracy: 57.6 -> 57.8No clear effect
add token recycling (ablation)VFlowOpt · POPE (retain 25%)accuracy: 86.8 -> 89.1Improved
add progressive pruning (three stages) (ablation)VFlowOpt · MMStar (retain 25%)accuracy: 56.0 -> 57.8Improved
optimize on last-token representation instead of mean pooling or first token (ablation)VFlowOpt Bayesian optimization · MMStar (retain 25%)accuracy: 56.1 (mean pooling), 54.2 (first token) -> 57.8Improved
choose optimization data from MathV360K-Geometry3K instead of random samples (ablation)VFlowOpt · MMStar and othersaccuracy: 57.8 -> 57.8No clear effect
prune 90% of visual tokens with VFlowOptLLaVA-OneVision-7B · MME efficiencylatency, KV cache, performance: latency 1040.1 -> 272.1 ms; KV cache -89.0%; MME 1581 -> 1464 (-7.4%)Mixed
prune 50% of visual tokens with VFlowOptLLaVA-OneVision-7B · MMEscore: 1581 -> 1591 (+0.6%)Improved
apply VFlowOpt at 25% retention to video benchmarksLLaVA-OneVision-7B · SeedBench video and VideoMMErelative avg performance (%): 99.0% (VisionZip) -> 100%Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy