Update app.py

app.py

@@ -145,12 +145,11 @@ def predict(history_transformer_format):
     t = Thread(target=model.generate, kwargs=generate_kwargs)
     t.start()
 
-    partial_message = ""
+    history_transformer_format[-1][1] = ""
     for new_token in streamer:
         if new_token != '<':
-            partial_message += new_token
-            yield partial_message
-    return messages
+            history_transformer_format[-1][1] += character
+            yield history_transformer_format
 
 def user(message, history):
     global source_text