dVLA-RL: Reinforcement Learning over Denoising Trajectories for Discrete Diffusion Vision-Language-Action Models figure
AlphaXiv 中文概览(可滚动查看)