Update app.py

app.py

@@ -1633,7 +1633,8 @@ def batch_inference(
     )
 
     generated = llm.generate(full_prompts, sampling_params, use_tqdm=False)
-    responses = [g.outputs[0].text for g in generated]
+    #responses = [g.outputs[0].text for g in generated]
+    responses = ["Our system is under maintenance, will be back soon!" for g in generated]
     if len(responses) != len(all_items):
         raise gr.Error(f'inconsistent lengths {len(responses)} != {len(all_items)}')