Update README.md

Added some advice about dealing with the model speaking as the user or out of turn during group chats.

README.md

@@ -160,6 +160,9 @@ If you save this as a .json file, you can import it directly into Silly Tavern.
     "name": "New Dawn Llama 3.1 70B"
 }
 ```
+**NOTE:** If you have trouble with this model speaking as the user or other characters out of turn during group chats, try setting 
+```<|start_header_id|>assistant - {{char}} ONLY<|end_header_id|>``` as the last_output_sequence, which is displayed in the SillyTavern GUI as "Last Assistant Prefix."
+This version definitely has more of a tendency to speak out of turn as compared to v1.0 that was based solely on Llama 3.
 
 ### Instruct Formats
 Use the Llama 3 instruct format. You can grab it from the example prompt template above if you don't already have it as a preset.