empty cache after generating an image

I found this workaround useful for using the interface locally.
Since i have only a 6GB GPU, i wasn't able to generate multiple images: after empty the cache i was able to generate as many image as i wanted.

I thought that this little trick can be useful for user with modest hardware like me that want to use the code locally.

Hope it helps!

Happy coding :)

app.py

@@ -184,6 +184,7 @@ def inference(
             "sampler": sampler,
         },
     )
+    torch.cuda.empty_cache()
 
     return out_image["images"][0], gr.update(visible=True), gr.update(visible=True), my_seed
         
@@ -274,7 +275,7 @@ with gr.Blocks(css=css) as app_with_history:
         app.render()
     with gr.Tab("Past generations"):
         user_history.render()
-
+    
 app_with_history.queue(max_size=20,api_open=False )
 
 if __name__ == "__main__":