Update app.py

app.py

@@ -1,25 +1,24 @@
 import gradio as gr
 import numpy as np
 import random
-import spaces # <--- UNCOMMENTED
+import spaces 
 from diffusers import DiffusionPipeline
 import torch
 
-device = "cuda" if torch.cuda.is_available() else "cpu"
+# 1. Load the model on CPU first (Safe for startup)
 model_repo_id = "stabilityai/sdxl-turbo" 
+pipe = DiffusionPipeline.from_pretrained(
+    model_repo_id, 
+    torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32
+)
 
-if torch.cuda.is_available():
-    torch_dtype = torch.float16
-else:
-    torch_dtype = torch.float32
+# NOTE: We do NOT move to cuda or enable xformers here. 
+# That would cause the crash you saw.
 
-pipe = DiffusionPipeline.from_pretrained(model_repo_id, torch_dtype=torch_dtype)
-pipe = pipe.to(device)
-pipe.enable_xformers_memory_efficient_attention() # <--- ADD THIS LINE
 MAX_SEED = np.iinfo(np.int32).max
 MAX_IMAGE_SIZE = 1024
 
-@spaces.GPU # <--- UNCOMMENTED: This gives you the H200 GPU power
+@spaces.GPU(duration=60) # Magic starts here
 def infer(
     prompt,
     negative_prompt,
@@ -33,7 +32,15 @@ def infer(
 ):
     if randomize_seed:
         seed = random.randint(0, MAX_SEED)
-    generator = torch.Generator().manual_seed(seed)
+    
+    # 2. Move to GPU and enable speed boosts ONLY when the function runs
+    pipe.to("cuda")
+    try:
+        pipe.enable_xformers_memory_efficient_attention()
+    except Exception:
+        pass # Fallback if xformers isn't needed
+        
+    generator = torch.Generator("cuda").manual_seed(seed)
     
     image = pipe(
         prompt=prompt,
@@ -44,6 +51,7 @@ def infer(
         height=height,
         generator=generator,
     ).images[0]
+    
     return image, seed
 
 # UI Styling
@@ -51,39 +59,32 @@ css = """#col-container { margin: 0 auto; max-width: 640px; }"""
 
 with gr.Blocks(css=css) as demo:
     with gr.Column(elem_id="col-container"):
-        gr.Markdown(" # CodeIgnite Image Engine (SDXL-Turbo)")
+        gr.Markdown(" # CodeIgnite Image Engine")
         
         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")
+            prompt = gr.Text(label="Prompt", show_label=False, placeholder="Enter your prompt", container=False)
+            run_button = gr.Button("Run", 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=True)
+            negative_prompt = gr.Text(label="Negative prompt", placeholder="Optional")
             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=512) # Lowered for speed
+                width = gr.Slider(label="Width", minimum=256, maximum=MAX_IMAGE_SIZE, step=32, value=512)
                 height = gr.Slider(label="Height", minimum=256, maximum=MAX_IMAGE_SIZE, step=32, value=512)
             
             with gr.Row():
                 guidance_scale = gr.Slider(label="Guidance scale", minimum=0.0, maximum=10.0, step=0.1, value=0.0)
-                num_inference_steps = gr.Slider(label="Inference steps", minimum=1, maximum=4, step=1, value=2)
+                num_inference_steps = gr.Slider(label="Steps", minimum=1, maximum=4, step=1, value=2)
 
-    # This 'api_name' is CRITICAL for your WhatsApp Bot to find the function
     run_button.click(
         fn=infer,
         inputs=[prompt, negative_prompt, seed, randomize_seed, width, height, guidance_scale, num_inference_steps],
         outputs=[result, seed],
-        api_name="predict" 
+        api_name="predict"
     )
 
 if __name__ == "__main__":