@arankomatsuzaki There is no acknowledgment or perhaps even awareness that the 3.5 model they tested was trained with RLHF, not just SSL. RLHF models, unlike base models, write like a PR department…

Posters: repligate. repligate

5 likes recorded by the source corpus.

Topics: AI & LLMs

View Post: https://x.com/repligate/status/1650431447143153666