Update app.py

app.py

@@ -114,7 +114,8 @@ def generate(
             else:
                 adapter_name2 = cached_loras[lora_key2]
             pipe.set_adapters([adapter_name1, adapter_name2], adapter_weights=[lora_scale, lora_scale2])
-    
+
+    enable_model_cpu_offload()
     pipe.to(device)
     generator = torch.Generator().manual_seed(seed)