<ul><li>The Weak-to-Strong Diffusion (W2SD) framework is proposed to reduce the gap between generated outputs and real data in diffusion generative models.</li><li>W2SD utilizes the estimated difference between weak and strong models to bridge the gap and align latent variables with the real data distribution.</li><li>The W2SD framework is highly flexible and applicable to various model pairs and modalities, achieving state-of-the-art performance.</li><li>Experiments demonstrate significant improvements in human preference, aesthetic quality, and prompt adherence with W2SD.</li></ul>

Weak-to-Strong Diffusion with Reflection

Discover more