Isn’t RLHF also instruction finetuning? You mean supervised instruction tuned?

→ View original post on X — @rasbt