Update README.md

README.md

@@ -145,4 +145,4 @@ while True:
 
 ## Known issues
 
-Inconsistent responses, including occasional nonsensical or strange answers. The AI may mistakenly identify itself as the user when asked about its identity, attributing the user's name to itself. The AI was only trained on 10-MB of conversational data, for testing purposes, bigger models will include more data. One of the bigger issues is that the AI ends the conversation. And starts generating a random personality for itself and making a whole new conversation. Will be fixed in bigger models.
+Inconsistent responses, including occasional nonsensical or strange answers. The AI may mistakenly identify itself as the user when asked about its identity, attributing the user's name to itself. The AI was only trained on 10-MB of conversational data, for testing purposes, bigger models will include more data. One of the bigger issues is that the AI ends the conversation. And starts generating a random personality for itself and making a whole new conversation. Will be fixed in bigger models. The AI sometimes doesn't play as the stated personality.