Update app.py

app.py

@@ -1,154 +1,92 @@
 import gradio as gr
-import numpy as np
-import random
-
-import spaces #[uncomment to use ZeroGPU]
-from diffusers import StableDiffusionXLPipeline, EulerAncestralDiscreteScheduler
 import torch
-
-device = "cuda" if torch.cuda.is_available() else "cpu"
-model_repo_id = "anon4ik/noobaiXLNAIXL_epsilonPred05Version_diffusers"  # Replace to the model you would like to use
-
-pipe = StableDiffusionXLPipeline.from_pretrained(model_repo_id, torch_dtype=torch.float16)
-pipe.scheduler = EulerAncestralDiscreteScheduler.from_config(pipe.scheduler.config)
-pipe.scheduler.register_to_config(
-    prediction_type="v_prediction",
-    rescale_betas_zero_snr=True,
-)
-pipe = pipe.to(device)
-
-MAX_SEED = np.iinfo(np.int32).max
-MAX_IMAGE_SIZE = 1024
-
-
-@spaces.GPU #[uncomment to use ZeroGPU]
-def infer(
-    prompt,
-    negative_prompt,
-    seed,
-    randomize_seed,
-    width,
-    height,
-    guidance_scale,
-    num_inference_steps,
-    progress=gr.Progress(track_tqdm=True),
-):
-    if randomize_seed:
-        seed = random.randint(0, MAX_SEED)
-
-    generator = torch.Generator().manual_seed(seed)
-
-    image = pipe(
-        prompt=prompt,
-        negative_prompt=negative_prompt,
-        guidance_scale=guidance_scale,
-        num_inference_steps=num_inference_steps,
-        width=width,
-        height=height,
-        generator=generator,
-    ).images[0]
-
-    return image, seed
-
-
-examples = [
-    "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k",
-    "An astronaut riding a green horse",
-    "A delicious ceviche cheesecake slice",
-]
-
-css = """
-#col-container {
-    margin: 0 auto;
-    max-width: 640px;
+from diffusers import StableDiffusionXLPipeline, EulerDiscreteScheduler
+from huggingface_hub import hf_hub_download
+from safetensors.torch import load_file
+import spaces
+from PIL import Image
+
+SAFETY_CHECKER = False
+
+# Constants
+base = "stabilityai/stable-diffusion-xl-base-1.0"
+repo = "advokat/noobaiXLNAIXL_epsilonPred075"
+checkpoints = {
+    "1-Step" : ["noobaiXLNAIXL_epsilonPred075.safetensors", 1],
 }
-"""
+loaded = None
 
-with gr.Blocks(css=css) as demo:
-    with gr.Column(elem_id="col-container"):
-        gr.Markdown(" # Text-to-Image Gradio Template")
+# Ensure model and scheduler are initialized in GPU-enabled function
+if torch.cuda.is_available():
+    pipe = StableDiffusionXLPipeline.from_pretrained(base, torch_dtype=torch.float16, variant="fp16").to("cuda")
 
-        with gr.Row():
-            prompt = gr.Text(
-                label="Prompt",
-                show_label=False,
-                max_lines=1,
-                placeholder="Enter your prompt",
-                container=False,
-            )
-
-            run_button = gr.Button("Run", scale=0, variant="primary")
-
-        result = gr.Image(label="Result", show_label=False)
-
-        with gr.Accordion("Advanced Settings", open=False):
-            negative_prompt = gr.Text(
-                label="Negative prompt",
-                max_lines=1,
-                placeholder="Enter a negative prompt",
-                visible=False,
-            )
-
-            seed = gr.Slider(
-                label="Seed",
-                minimum=0,
-                maximum=MAX_SEED,
-                step=1,
-                value=0,
-            )
-
-            randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
-
-            with gr.Row():
-                width = gr.Slider(
-                    label="Width",
-                    minimum=256,
-                    maximum=MAX_IMAGE_SIZE,
-                    step=32,
-                    value=1024,  # Replace with defaults that work for your model
-                )
-
-                height = gr.Slider(
-                    label="Height",
-                    minimum=256,
-                    maximum=MAX_IMAGE_SIZE,
-                    step=32,
-                    value=1024,  # Replace with defaults that work for your model
-                )
-
-            with gr.Row():
-                guidance_scale = gr.Slider(
-                    label="Guidance scale",
-                    minimum=0.0,
-                    maximum=10.0,
-                    step=0.1,
-                    value=0.0,  # Replace with defaults that work for your model
-                )
-
-                num_inference_steps = gr.Slider(
-                    label="Number of inference steps",
-                    minimum=1,
-                    maximum=50,
-                    step=1,
-                    value=2,  # Replace with defaults that work for your model
-                )
-
-        gr.Examples(examples=examples, inputs=[prompt])
-    gr.on(
-        triggers=[run_button.click, prompt.submit],
-        fn=infer,
-        inputs=[
-            prompt,
-            negative_prompt,
-            seed,
-            randomize_seed,
-            width,
-            height,
-            guidance_scale,
-            num_inference_steps,
-        ],
-        outputs=[result, seed],
+if SAFETY_CHECKER:
+    from safety_checker import StableDiffusionSafetyChecker
+    from transformers import CLIPFeatureExtractor
+
+    safety_checker = StableDiffusionSafetyChecker.from_pretrained(
+        "CompVis/stable-diffusion-safety-checker"
+    ).to("cuda")
+    feature_extractor = CLIPFeatureExtractor.from_pretrained(
+        "openai/clip-vit-base-patch32"
     )
 
-if __name__ == "__main__":
-    demo.launch()
+    def check_nsfw_images(
+        images: list[Image.Image],
+    ) -> tuple[list[Image.Image], list[bool]]:
+        safety_checker_input = feature_extractor(images, return_tensors="pt").to("cuda")
+        has_nsfw_concepts = safety_checker(
+            images=[images],
+            clip_input=safety_checker_input.pixel_values.to("cuda")
+        )
+
+        return images, has_nsfw_concepts
+
+# Function 
+@spaces.GPU(enable_queue=True)
+def generate_image(prompt, ckpt):
+    global loaded
+    print(prompt, ckpt)
+
+    checkpoint = checkpoints[ckpt][0]
+    num_inference_steps = checkpoints[ckpt][1]
+
+    if loaded != num_inference_steps:
+        pipe.scheduler = EulerDiscreteScheduler.from_config(pipe.scheduler.config, timestep_spacing="trailing", prediction_type="sample" if num_inference_steps==1 else "epsilon")
+        pipe.unet.load_state_dict(load_file(hf_hub_download(repo, checkpoint), device="cuda"))
+        loaded = num_inference_steps
+        
+    results = pipe(prompt, num_inference_steps=num_inference_steps, guidance_scale=0)
+
+    if SAFETY_CHECKER:
+        images, has_nsfw_concepts = check_nsfw_images(results.images)
+        if any(has_nsfw_concepts):
+            gr.Warning("NSFW content detected.")
+            return Image.new("RGB", (512, 512))
+        return images[0]
+    return results.images[0]
+
+
+
+# Gradio Interface
+
+with gr.Blocks(css="style.css") as demo:
+    gr.HTML("<h1><center>SDXL-Lightning ⚡</center></h1>")
+    gr.HTML("<p><center>Lightning-fast text-to-image generation</center></p><p><center><a href='https://huggingface.co/ByteDance/SDXL-Lightning'>https://huggingface.co/ByteDance/SDXL-Lightning</a></center></p>")
+    with gr.Group():
+        with gr.Row():
+            prompt = gr.Textbox(label='Enter your prompt (English)', scale=8)
+            ckpt = gr.Dropdown(label='Select inference steps',choices=['1-Step', '2-Step', '4-Step', '8-Step'], value='4-Step', interactive=True)
+            submit = gr.Button(scale=1, variant='primary')
+    img = gr.Image(label='SDXL-Lightning Generated Image')
+
+    prompt.submit(fn=generate_image,
+                 inputs=[prompt, ckpt],
+                 outputs=img,
+                 )
+    submit.click(fn=generate_image,
+                 inputs=[prompt, ckpt],
+                 outputs=img,
+                 )
+    
+demo.queue().launch()