Originally published here by Anastasia Berezniova
At datarockets, we have a small tradition of greeting teammates with a Happy Friday post in th...
For further actions, you may consider blocking this person and/or reporting abuse
👌🏾 Could I ask why GAN and not Diffusion Models?
Diffusion models are considered to take more time to train than GANs, we had neither a lot of GPU power nor time, so went for what we thought would let us wait less.
But in the end, the last version of our PGGAN still turned out pretty heavy (for edu project of course) and took around 12 hours to train.