app.py

from diffusers import DiffusionPipeline
import torch
import PIL.Image
import gradio as gr
import random
import numpy as np

pipeline = DiffusionPipeline.from_pretrained("1aurent/ddpm-mnist")

def predict(steps, seed):
    generator = torch.manual_seed(seed)
    for i in range(1,steps):
        yield pipeline(generator=generator, num_inference_steps=i).images[0]

random_seed = random.randint(0, 2147483647)
gr.Interface(
    predict,
    inputs=[
        gr.inputs.Slider(1, 100, label='Inference Steps', default=5, step=1),
        gr.inputs.Slider(0, 2147483647, label='Seed', default=random_seed, step=1),
    ],
    outputs=gr.Image(shape=[28,28], type="pil", elem_id="output_image"),
    css="#output_image{width: 256px}",
    title="Unconditional butterflies",
    description="A DDPM scheduler and UNet model trained on the MNIST dataset for unconditional image generation.",
).queue().launch()