SDPO: Importance-Sampled Direct Preference Optimization for Stable Diffusion Training

Re-assign community