@DavidSHolz @YeshuaisSavior it empirically seems & makes sense that RLHF steers towards agreeableness/sycophancy while constitutional RLAIF steers towards a character that behaves with presumed…

Posters: repligate. repligate

3 likes recorded by the source corpus.

Topics: AI & LLMs

View Post: https://x.com/repligate/status/1716638993662500998