Depends on the LLM and its training data. Will have to think more about the second question.

→ View original post on X — @mattshumer_