TY  - RPRT
TI  - From Seeing to Thinking: Decoupling Perception and Reasoning Improves Post-Training of Vision-Language Models
AU  - Juncheng Wu
AU  - Hardy Chen
AU  - Haoqin Tu
AU  - Xianfeng Tang
AU  - Freda Shi
AU  - Hui Liu
AU  - Hanqing Lu
AU  - Cihang Xie
AU  - Yuyin Zhou
PY  - 2026
UR  - https://arxiv.org/abs/2605.20177
ID  - 2605.20177
ER  -