from diffusers import StableDiffusionPipeline import torch model_id = "runwayml/stable-diffusion-v1-5" pipe = StableDiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16) pipe = pipe.to("cuda") prompt = "a photo of an astronaut riding a horse on mars" image = pipe(prompt).images[0] image.save("astronaut_rides_horse.png")