TY  - RPRT
TI  - AdvFidelity: Unifying Preference Alignment and Few-Step Image Generation within Reinforcement Learning
AU  - Sicong Kang
AU  - Fangao Zeng
AU  - Yuhan Li
AU  - Hao Zhou
AU  - Mengfei Xu
AU  - Wenxiang Shang
AU  - Wei Li
AU  - Bingbing Ni
AU  - Pipei Huang
PY  - 2026
UR  - https://arxiv.org/abs/2610.29380
ID  - 2610.29380
ER  -