WAM-RL: World-Action Model Reinforcement Learning with Reconstruction Rewards and Online Video SFT figure
AlphaXiv 中文概览(可滚动查看)