TDM-R1: Reinforcing Few-Step Diffusion Models with Non-Differentiable Reward Paper • 2603.07700 • Published Mar 8 • 13
