Update app.py

app.py

@@ -25,10 +25,8 @@ def respond(message, history):
     if cfg.streaming:
         # Call stream response and stream output
         stream = vq.submit_query_streaming(message)
-        outputs = ""
         for output in stream:
-            outputs += output
-            yield outputs
+            yield output
     else:
         # Call non-stream response and return message output
         response = vq.submit_query(message)
@@ -74,11 +72,19 @@ with gr.Blocks(css=bot_css) as demo:
     def bot(history):
         message = history[-1][0]
         bot_message = respond(message, history)
-        history[-1][1] = next(bot_message)
-        return history
+
+        if cfg.streaming:
+            full_response = ""
+            for chunk in bot_message:
+                full_response += chunk
+                history[-1][1] = full_response
+                yield history
+        else:
+            history[-1][1] = next(bot_message)
+            yield history
 
     msg.submit(user, [msg, chatbot], [msg, chatbot], queue=False).then(
-        bot, chatbot, chatbot
+        bot, chatbot, chatbot, api_name="bot_response"
     )
 
     clear.click(lambda: None, None, chatbot, queue=False)