People should just try base models, or models where you have access to prefills. I guess so many who use LLMs nowadays have never actually experimented with a t

People should just try base models, or models where you have access to prefills. I guess so many who use LLMs nowadays have never actually experimented with a text completion endpoint locally/over API.