Custom Diffusion + StableDiffusion Training UI

[cibernicola]

An integration would be really interesting :):

Custom Diffusion allows you to fine-tune text-to-image diffusion models, such as Stable Diffusion, given a few images of a new concept (~4-20). We fine-tune only a subset of model parameters, namely key and value projection matrices, in the cross-attention layers and the modifier token used to represent the object. This also reduces the extra storage for each additional concept to 75MB. Our method also allows you to use a combination of concepts. There's still limitations on which compositions work. For more analysis please refer to our website.

More info: https://huggingface.co/spaces/nupurkmr9/custom-diffusion