from diffusers import OnnxStableDiffusionPipeline cfg = 8 steps = 22 height = 512 width = 512 prompt = "an astronaut eating a hamburger" pipe = OnnxStableDiffusionPipeline.from_pretrained("../models/stable-diffusion-onnx-v1-5", provider="DmlExecutionProvider", safety_checker=None) image = pipe(prompt, height, width, num_inference_steps=steps, guidance_scale=cfg).images[0] image.save("../test.png")