AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

WinTok: A Win-Win Hybrid Tokenizer via Decomposing Visual Understanding and Generation with Transferable Tokens

Yiwei Guo, Shaobin Zhuang, Zhipeng Huang, Canmiao Fu, Chen Li, Jing LYU, Yali Wang

ECCV 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Interleaved image-text generation with MLLMs · 1 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
increase learnable tokens from 32 to 256 (ablation)WinTok · ImageNet-1KTop-1 accuracy: 78.8 -> 82.0Improved
increase learnable tokens from 32 to 256 (ablation)WinTok · ImageNet-1KrFID: 0.55 -> 0.42Improved
increase learnable tokens from 32 to 256 (ablation)WinTok · MME-PMME-P: 1248.0 -> 1453.0Improved
use SigLIP2-So400M teacher instead of CLIP-L/14 (ablation)WinTok · MME-PMME-P: 1328.4 -> 1453.0Improved
use SigLIP2-So400M teacher instead of DINOv2-L (ablation)WinTok · MME-PMME-P: 1274.5 -> 1453.0Improved
use SigLIP2 teacher instead of CLIP-L/14 (ablation)WinTok · MMBenchaccuracy: 58.1 -> 64.7Improved
use larger ViT-XL decoder instead of ViT-B (ablation)WinTok · ImageNet-1KrFID: 0.60 -> 0.54Improved
use larger ViT-XL decoder instead of ViT-B (ablation)WinTok · generationgFID: 16.51 -> 14.72Improved
swap roles of learnable and pixel tokens (LoseTok) (ablation)WinTok · reconstructionrFIDHurt
add learnable semantic tokens with distillation vs UniTokWinTok · ImageNet-1KTop-1 accuracy: 70.8 -> 82.0Improved
use 256 instead of 64 semantic tokens in LLM (ablation)WinTok · MMBenchaccuracy: 60.2 -> 74.9Improved
use 256 instead of 64 semantic tokens in LLM (ablation)WinTok · POPEaccuracy: 84.1 -> 86.5Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 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