add interactive mode

app.py

@@ -36,7 +36,7 @@ def chat(user_input, max_new_tokens=512):
     thread = Thread(target=generate)
     thread.start()
 
-    start_idx = len(inputs)
+    start_idx = len("SYSTEM") + len(messages[0]["content"]) + len("HUMAN") + len(user_input) + len("ASSISTANT")
     generated_text = ""
     for new_text in streamer:
         generated_text += new_text