arXiv preprint / cs.CV

VideoAWR: Reinforcing Video Flow Models with Decoupled-Timestep Advantage-Weighted Rectification

作者：Zheming Fu, Wenbin Ning, Ruizhe He, Wei Shang, Xiaoxiao Ma, Lianrui Mu, Chang Liu, Lei Wang, Siming Fu · 会议/期刊：arXiv preprint · 方向：cs.CV · 发布日期：2026-10-12

2026-10-12视觉感知生成模型
新标签打开 AlphaXiv新标签打开 Code
VideoAWR: Reinforcing Video Flow Models with Decoupled-Timestep Advantage-Weighted Rectification figure
AlphaXiv 中文概览（可滚动查看）