Comment by hubraumhugo

Just updated my HN Wrapped project with it and it does well on my totally unscientific LLM humor benchmark: https://hn-wrapped.kadoa.com