AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Can Pc a Be Run at Any of the Intermediate Layers of the Generator?
We run pc a on very early layers of the generator because that is computatially feasible. For big gan, it gets really expensive to run p at some of the later layers. But i would say in general, this is still fast. We'd love to see people use our method on all kinds of different generators and try to find different kinds of interesting knobs and tweaks. And there's the video i was mentioning, and it's showing off some soft ware that exposes the controls. You can see the output and what not.