Update app.py

app.py

@@ -3,17 +3,20 @@ import numpy as np
 import random
 from diffusers import DiffusionPipeline
 import torch
+from optimum.intel import ipex
 
-device = "cuda" if torch.cuda.is_available() else "cpu"
+# Use Intel Extension for PyTorch for CPU optimization
+device = "cpu"
 
-if torch.cuda.is_available():
-    torch.cuda.max_memory_allocated(device=device)
-    pipe = DiffusionPipeline.from_pretrained("stabilityai/sdxl-turbo", torch_dtype=torch.float16, variant="fp16", use_safetensors=True)
-    pipe.enable_xformers_memory_efficient_attention()
-    pipe = pipe.to(device)
-else: 
-    pipe = DiffusionPipeline.from_pretrained("stabilityai/sdxl-turbo", use_safetensors=True)
-    pipe = pipe.to(device)
+# Load the pipeline with optimizations for CPU inference
+pipe = DiffusionPipeline.from_pretrained(
+    "stabilityai/sdxl-turbo", 
+    use_safetensors=True
+)
+pipe = pipe.to(device)
+
+# Optimize the pipeline using Intel Extension for PyTorch
+ipex.optimize(pipe.unet, dtype=torch.float32)  # Optimized for Intel CPUs
 
 MAX_SEED = np.iinfo(np.int32).max
 MAX_IMAGE_SIZE = 1024
@@ -22,26 +25,28 @@ def infer(prompt_part1, color, dress_type, front_design, back_design, prompt_par
     if randomize_seed:
         seed = random.randint(0, MAX_SEED)
     
-    generator = torch.Generator().manual_seed(seed)
+    generator = torch.Generator(device=device).manual_seed(seed)
     
+    # Front view prompt generation and inference
     front_prompt = f"front view of {prompt_part1} {color} colored plain {dress_type} with {front_design} design, {prompt_part5}"
     front_image = pipe(
-        prompt=front_prompt, 
+        prompt=front_prompt,
         negative_prompt=negative_prompt,
-        guidance_scale=guidance_scale, 
-        num_inference_steps=num_inference_steps, 
-        width=width, 
+        guidance_scale=guidance_scale,
+        num_inference_steps=num_inference_steps,
+        width=width,
         height=height,
         generator=generator
     ).images[0]
     
+    # Back view prompt generation and inference
     back_prompt = f"back view of {prompt_part1} {color} colored plain {dress_type} with {back_design} design, {prompt_part5}"
     back_image = pipe(
-        prompt=back_prompt, 
+        prompt=back_prompt,
         negative_prompt=negative_prompt,
-        guidance_scale=guidance_scale, 
-        num_inference_steps=num_inference_steps, 
-        width=width, 
+        guidance_scale=guidance_scale,
+        num_inference_steps=num_inference_steps,
+        width=width,
         height=height,
         generator=generator
     ).images[0]
@@ -61,17 +66,12 @@ css = """
 }
 """
 
-if torch.cuda.is_available():
-    power_device = "GPU"
-else:
-    power_device = "CPU"
-
 with gr.Blocks(css=css) as demo:
     
     with gr.Column(elem_id="col-container"):
         gr.Markdown(f"""
         # Text-to-Image Gradio Template
-        Currently running on {power_device}.
+        Currently running on CPU (Optimized).
         """)
         
         with gr.Row():
@@ -179,15 +179,15 @@ with gr.Blocks(css=css) as demo:
                     minimum=0.0,
                     maximum=10.0,
                     step=0.1,
-                    value=0.0,
+                    value=7.5,  # Default value optimized for accuracy and speed
                 )
                 
                 num_inference_steps = gr.Slider(
                     label="Number of inference steps",
                     minimum=1,
-                    maximum=12,
+                    maximum=12,  # Reduced steps for faster execution
                     step=1,
-                    value=2,
+                    value=8,  # Balanced between speed and quality
                 )
         
         gr.Examples(