Make Diffusion Model's GAN's again

Hello,

We're a community-driven organization researching on Consistency Models. We're trying to drastically reduce the number of inference steps for diffusion models so that they can become as fast as GANs.

In a first step we've added OpenAI's released checkpoints to 🧨 Diffusers which you can access here.

In a next step, we're looking into training consistency models on larger text-conditioned datasets.