Rescaling Intermediate Features Makes Trained Consistency Models Perform Better.

Junyi Zhu, Zinan Lin, Enshu Liu, Xuefei Ning, Matthew B. Blaschko

Browse the full ICLR paper archive.