Update app.py

app.py

@@ -19,25 +19,23 @@ def respond(
     top_p,
 ):
     messages = [{"role": "system", "content": system_message}]
+    message = f"<s> {system_message} [INST] {message} [/INST]"
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+    
 
-    messages.append({"role": "user", "content": message})
+    final = ""
 
-    response = ""
-
-    response  =   p(messages,
+    for response   in  p(message,
         max_new_tokens=20,
         do_sample=True,
         temperature=1,
     )
 
-    return response
+        text = response["generated_text"]
+        final += text
+        yield text
 
+    return final
 
 """
 For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface