import torch
from diffusers import DiffusionPipeline
# switch to "mps" for apple devices
pipe = DiffusionPipeline.from_pretrained("greentree/SDXL-olive-optimized", dtype=torch.bfloat16, device_map="cuda")
prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"
image = pipe(prompt).images[0]Model Stable Diffusion XL 1.0 Base optimized using Microsoft Olive (https://github.com/microsoft/Olive). Provides massively increased generation speed on my AMD RX 7900 XT on Windows 11 for images of size 1024*1024. (~10s/it --> ~3.80it/s)
Warning: Requires ONNX Runtime, so this will not work interchangably with any other custom model. I hope all these downloads weren't just like four people trying in vain.
Warning 2: Outputs from this model will contain some noise and may need to be refined using another model. I'm commited to updating this if the optimization is improved.
Credit for the original model to StabilityAI: https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0
Other models I've uploaded: https://huggingface.co/greentree/SDXL-Refiner-olive-optimized, https://huggingface.co/greentree/Dreamshaper-XL-olive-optimized, https://huggingface.co/greentree/playground-v2-olive-optimized
Also see a collection of many more models like this by https://huggingface.co/softwareweaver
- Downloads last month
- 532