Our paper “Information Theoretic Learning for Diffusion Models with Warm Start” has been accepted by NeurIPS 2025!