AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

TokenPacker: Efficient Visual Projector for Multimodal LLM

Wentong Li, Yuqian Yuan, Jian Liu, Dongqi Tang, Song Wang, Jie Qin, Jianke Zhu, Lei Zhang

IJCV 2025 · Vision-language and multimodal › Multimodal LLMs · approach: Instruction-tuned multimodal assistants · 165 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
add region-to-point injection module (ablation)2x downsampled baseline projector · VQAv2accuracy: 76.4 -> 77.5Improved
add region-to-point injection module (ablation)2x downsampled baseline projector · TextVQAaccuracy: 55.3 -> 56.5Improved
use learnable query instead of downsampled feature query (ablation)TokenPacker injection · GQAaccuracy: 61.6 -> 59.8Hurt
add multi-level features as keys and values (ablation)TokenPacker injection · TextVQAaccuracy: 56.5 -> 57.2Improved
add dynamic image slicing (ablation)TokenPacker · TextVQAaccuracy: 57.2 -> 64.0Improved
add separator tokens (comma and newline) between grids (ablation)TokenPacker with image slicing · TextVQAaccuracy: 58.3 -> 64.0Improved
use dynamic slicing instead of FixedSplit (ablation)TokenPacker-HD · OCRBenchscore: 327 -> 336Improved
use dynamic slicing instead of AdaptiveSplit (ablation)TokenPacker-HD · TextVQAaccuracy: 63.4 -> 64.0Improved
use multi-level layers 12-16-22-23 instead of single layer 23 (ablation)TokenPacker injection · TextVQAaccuracy: 56.5 -> 57.2Improved
use shallower multi-level layers 8-12-22-23 (ablation)TokenPacker injection · VQAv2accuracy: 77.9 -> 76.8Hurt
replace MLP with TokenPacker at 144 tokensLLaVA-1.5 Vicuna-7B · Avg of six benchmarksaverage score: 62.0 -> 62.8Improved
replace LDP-v2 with TokenPacker at 144 tokens (ablation)LLaVA-1.5 Vicuna-7B · Avg of six benchmarksaverage score: 61.2 -> 62.8Improved

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