Update app.py

app.py

@@ -60,22 +60,26 @@ def generate(
             
         if not use_img2img:
             pipe = DiffusionPipeline.from_pretrained(model, torch_dtype=torch.float16) 
-            pipe.enable_sequential_cpu_offload()
+                pipe.enable_model_cpu_offload()
+                pipe.unet = torch.compile(pipe.unet, mode="reduce-overhead", fullgraph=True)
 
             if use_vae:  
                 vae = AutoencoderKL.from_pretrained(vaecall, torch_dtype=torch.float16)
                 pipe = DiffusionPipeline.from_pretrained(model, vae=vae, torch_dtype=torch.float16)
-                pipe.enable_sequential_cpu_offload()
+                pipe.enable_model_cpu_offload()
+                pipe.unet = torch.compile(pipe.unet, mode="reduce-overhead", fullgraph=True)
 
         if use_img2img:
             pipe = AutoPipelineForImage2Image.from_pretrained(model, torch_dtype=torch.float16)
-            pipe.enable_sequential_cpu_offload()
+            pipe.enable_model_cpu_offload()
+            pipe.unet = torch.compile(pipe.unet, mode="reduce-overhead", fullgraph=True)
             
             if use_vae:  
                 vae = AutoencoderKL.from_pretrained(vaecall, torch_dtype=torch.float16)
                 pipe = AutoPipelineForImage2Image.from_pretrained(model, vae=vae, torch_dtype=torch.float16)
-                pipe.enable_sequential_cpu_offload()
-
+                pipe.enable_model_cpu_offload()
+                pipe.unet = torch.compile(pipe.unet, mode="reduce-overhead", fullgraph=True)
+                
             response = requests.get(url)
             init_image = Image.open(BytesIO(response.content)).convert("RGB")
             init_image = init_image.resize((width, height))