2 ms·
Is it known that Bard is trained using RLHF. My guess was that it wasn't.
by YetAnotherNick 3y ago
Is it known that Bard is trained using RLHF. My guess was that it wasn't.
- ipsum2 3y agoI suspect GPT-4 has some magic that's not only the original RLHF using PPO. Either an augmented dataset or some enhanced version of RLHF.