check_app.py

import spaces
import torch
from diffusers import (
    FluxPipeline,
    StableDiffusion3Pipeline,
    PixArtSigmaPipeline,
    SanaPipeline,
    AuraFlowPipeline,
    Kandinsky3Pipeline,
    HunyuanDiTPipeline,
    LuminaText2ImgPipeline
)
import gradio as gr

cache_dir = '/workspace/hf_cache'

MODEL_CONFIGS = {
    "FLUX": {
        "repo_id": "black-forest-labs/FLUX.1-dev",
        "pipeline_class": FluxPipeline,
        "cache_dir": cache_dir,
    },
    "Stable Diffusion 3.5": {
        "repo_id": "stabilityai/stable-diffusion-3.5-large",
        "pipeline_class": StableDiffusion3Pipeline,
        "cache_dir": cache_dir,
    },
    "PixArt": {
        "repo_id": "PixArt-alpha/PixArt-Sigma-XL-2-1024-MS",
        "pipeline_class": PixArtSigmaPipeline,
        "cache_dir": cache_dir,
    },
    "SANA": {
        "repo_id": "Efficient-Large-Model/Sana_1600M_1024px_BF16_diffusers",
        "pipeline_class": SanaPipeline,
        "cache_dir": cache_dir,
    },
    "AuraFlow": {
        "repo_id": "fal/AuraFlow",
        "pipeline_class": AuraFlowPipeline,
        "cache_dir": cache_dir,
    },
    "Kandinsky": {
        "repo_id": "kandinsky-community/kandinsky-3",
        "pipeline_class": Kandinsky3Pipeline,
        "cache_dir": cache_dir,
    },
    "Hunyuan": {
        "repo_id": "Tencent-Hunyuan/HunyuanDiT-Diffusers",
        "pipeline_class": HunyuanDiTPipeline,
        "cache_dir": cache_dir,
    },
    "Lumina": {
        "repo_id": "Alpha-VLLM/Lumina-Next-SFT-diffusers",
        "pipeline_class": LuminaText2ImgPipeline,
        "cache_dir": cache_dir,
    }
}

def generate_image_with_progress(pipe, prompt, num_steps, guidance_scale=None, seed=None, progress=gr.Progress()):
    generator = None
    if seed is not None:
        generator = torch.Generator("cuda").manual_seed(seed)

    def callback(pipe, step_index, timestep, callback_kwargs):
        print(f" callback => {pipe}, {step_index}, {timestep}")
        if step_index is None:
            step_index = 0
        cur_prg = step_index / num_steps
        progress(cur_prg, desc=f"Step {step_index}/{num_steps}")
        return callback_kwargs

    if hasattr(pipe, "guidance_scale"):
        image = pipe(
            prompt,
            num_inference_steps=num_steps,
            guidance_scale=guidance_scale,
            callback_on_step_end=callback,
        ).images[0]
    else:
        image = pipe(
            prompt,
            num_inference_steps=num_steps,
            generator=generator,
            output_type="pil",
            callback_on_step_end=callback,
        ).images[0]

    return image

@spaces.GPU(duration=170)
def create_pipeline_logic(model_name, config):
    def start_process(prompt_text):
        print(f"starting {model_name}")
        progress = gr.Progress()
        num_steps = 30
        guidance_scale = 7.5  # Example guidance scale, can be adjusted per model
        seed = 42

        pipe_class = config["pipeline_class"]
        pipe = pipe_class.from_pretrained(
            config["repo_id"],
            #cache_dir=config["cache_dir"],
            torch_dtype=torch.bfloat16
        ).to("cuda")
        image = generate_image_with_progress(
            pipe, prompt_text, num_steps=num_steps, guidance_scale=guidance_scale, seed=seed, progress=progress
        )
        return f"Seed: {seed}", image

    return start_process


def main():
    with gr.Blocks() as app:
        gr.Markdown("# Dynamic Multiple Model Image Generation")

        prompt_text = gr.Textbox(label="Enter prompt")

        for model_name, config in MODEL_CONFIGS.items():
            with gr.Tab(model_name):
                button = gr.Button(f"Run {model_name}")
                output = gr.Textbox(label="Status")
                img = gr.Image(label=model_name, height=300)

                start_process = create_pipeline_logic(model_name, config)
                button.click(fn=start_process, inputs=[prompt_text], outputs=[output, img])

    app.launch()


if __name__ == "__main__":
    main()