Few-Step Generation via Data-Space Iteration
Flow matching has emerged as a scalable paradigm for training high-quality generative models, but sampling from the learned probability flow requires many network evaluations. Distillation can reduce this cost to one or a few evaluations; however, one-step generation often sacrifices quality, making few-step generation...
Shanchuan Lin, Yan-Song Peng, Fu-Yun Wang et al.
· 0 citations