6 ms·
I think people are misunderstanding reward functions and LLMs. LLMs don't actually have a reward system like some other ML models.
by xandrius 7mo ago
I think people are misunderstanding reward functions and LLMs.
LLMs don't actually have a reward system like some other ML models.
- storus 7mo agoThey are trained with one, and when you look at DPO you can say they contain an implicit one as well.