Tweeted By @danijarh
Excited to present Clockwork VAEs for video prediction!
— Danijar Hafner (@danijarh) February 23, 2021
Clockwork VAEs (CW-VAEs) leverage hierarchies of latent sequences, where higher levels tick slower. They learn long-term deps across 1000 frames, semantically separate content, and outperform strong video models.
👇 Thread pic.twitter.com/Cn11MAKjIB