Update app.py

app.py

@@ -1273,7 +1273,7 @@ def chat_response_stream_multiturn(
         assert len(gen) == 1, f'{gen}'
         item = next(iter(gen.values()))
         cur_out = item.outputs[0].text
-        cur_out = "Our system is under maintenance, will be back soon!"
+        #cur_out = "Our system is under maintenance, will be back soon!"
         if j >= max_tokens - 2:
             gr.Warning(f'The response hits limit of {max_tokens} tokens. Consider increase the max tokens parameter in the Additional Inputs.')
     
@@ -1633,8 +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 = ["Our system is under maintenance, will be back soon!" 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)}')