Dynamic Preference Optimization Boosts Low‑Cost Diffusion Sampling
D2PO transfers the Direct Preference Optimization framework, the core of RLHF, to diffusion model sampling. By jointly learning timestep schedules and classifier‑free guidance weights, it outperforms fixed‑schedule samplers, especially when the number of function evaluations is low, preserving fine texture while keeping global structure. This opens the door to cheaper, higher‑quality image generation.