Update app.py

app.py

@@ -14,10 +14,16 @@ def respond(message, history):
     if history:
         messages.extend(history)
     messages.append({"role": "user", "content": message})
-    response = client.chat_completion(
+    response = ""
+    client.chat_completion(
         messages,
         temperature = 0.1
+        stream = True
     )
+    for message in client.chat_completion:
+        token = messages.choices[0].delta.content
+        response += token
+        yield response
     return response['choices'][0]['message']['content'].strip()
     
 chatbot = gr.ChatInterface(respond, type = "messages")