Update gradio_demo.py

gradio_demo.py

@@ -95,15 +95,15 @@ def enable_efficient_attention():
             print(f"Xformers error: {e}")
             print("Falling back to sliced attention")
             # Use sliced attention for RTX 2070
-            unet.set_attention_slice_size(4)
-            vae.set_attention_slice_size(4)
+            # unet.set_attention_slice_size(4)
+            # vae.set_attention_slice_size(4)
             unet.set_attn_processor(AttnProcessor2_0())
             vae.set_attn_processor(AttnProcessor2_0())
     else:
         # Fallback for when xformers is not available
         print("Using sliced attention")
-        unet.set_attention_slice_size(4)
-        vae.set_attention_slice_size(4)
+        # unet.set_attention_slice_size(4)
+        # vae.set_attention_slice_size(4)
         unet.set_attn_processor(AttnProcessor2_0())
         vae.set_attn_processor(AttnProcessor2_0())