This is a model related to the "Picture of the Week" contest on
Stable Diffusion discord.
I try to make a model out of all the submission for people to continue enjoy the theme after the even, and see a little of their designs in other people's creations. The token stays "SDArt" and I balance the learning on the low side, so that it doesn't just replicate creations.
The total dataset is made of 47 pictures. It was trained on
Stable diffusion 2.1 768px. I used
EveryDream to do the training, 100 total repeat per picture. The pictures were tagged using the token "SDArt", and an arbitrary token I choose. The dataset is provided below, as well as a list of usernames and their corresponding token.
The recommended sampling is k_Euler_a or DPM++ 2M Karras on 20 steps, CFGS 7.5 .
The model is also available here in a version trained on 1.5 as a base.
This model can be used just like any other Stable Diffusion model. For more information,
please have a look at the
Stable Diffusion.
You can also export the model to
ONNX,
MPS and/or
FLAX/JAX.
1from diffusers import StableDiffusionPipeline
2import torch
3
4model_id = "Guizmus/SDArt_AliceInDiffusionLand768"
5pipe = StableDiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16)
6pipe = pipe.to("cuda")
7
8prompt = "SDArt pkg"
9image = pipe(prompt).images[0]
10
11image.save("./SDArt.png")
