Update demo.py

used pipe.enable_sequential_cpu_offload() for lower vram use.

demo.py

@@ -10,8 +10,8 @@ pipe = DiffusionPipeline.from_pretrained(
     torch_dtype=torch.bfloat16,
 )
 
-print("model initialized. Now moving to CUDA")
-pipe.to("cuda")
+print("model initialized.")
+pipe.enable_sequential_cpu_offload()
 
 print("Trying render now...")
 images = pipe("a misty Tokyo alley at night",num_inference_steps=30).images