Stable Diffusion Masterclass - Theory, Code & Application

Inference with a Diffusion Model

Now we understand how a diffusion model is trained to generate images. When we show up at a diffusion prompt on the web and specify what type of image we want the model to generate for us, however, what occurs at this point is not training, but rather, inference.

As we're familiar, using a network for inference purposes consists of passing data as input to the trained network and completing a forward pass to get the network's output.