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 · 2026

Read WinTok: A Win-Win Hybrid Tokenizer via Decomposing Visual Understanding and Generation with Transferable Tokens by Yiwei Guo, Shaobin Zhuang, Zhipeng Huang, Canmiao Fu, Chen Li, Jing LYU, Yali Wang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-18115.txt.

Description

Yiwei Guo and six colleagues present WinTok, a hybrid visual tokenizer splitting visual reading from making with shared tokens. Single token spaces strain to serve both semantic ideas and pixel rebuild, so WinTok adds tuned semantic tokens to pixel tokens, easing cross-task clash without dual tokenizers. A one-sided teaching step guides semantic tokens with pretrained vectors from any visual base model, lending strong sorting power. Across 10 benchmarks, WinTok improves rebuild, reading, and making, beating UniTok by 11.2 percent in classification accuracy with rFID 0.41 on only 50M open-source data.

Catalog record

CREATORYiwei Guo, Shaobin Zhuang, Zhipeng Huang, Canmiao Fu, Chen Li, Jing LYU, Yali Wang
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE47486
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.18115

Browse the full library · About the Tower of Babel