A Deep Dive Into Stable Diffusion and Other Leading Text-to-Image Models | HackerNoon
Briefly

Stable Diffusion is a family of 1B-parameter text-to-image models based on latent diffusion, trained on the large-scale paired text-image dataset LAION.
Stable Diffusion models, like v1-4 and v1-5, utilize advanced training settings including dropping text-conditioning to enhance classifier-free guidance sampling.
Read at Hackernoon
[
|
]