Update app.py

app.py

@@ -162,8 +162,8 @@ pipe.to("cuda")
 # pipe.load_lora_weights("TODO/TODO", adapter_name="ltx-lora")
 # pipe.set_adapters(["lrx-lora"], adapter_weights=[1.0])
 
-pipe.vae.to(memory_format=torch.channels_last)
 pipe.vae.decode = torch.compile(pipe.vae.decode, mode="max-autotune", fullgraph=True)
+pipe.vae = torch.compile(pipe.vae.decode, mode="reduce-overhead", fullgraph=True)
 
 @spaces.GPU(duration=120)
 @torch.inference_mode()