Re:OpenAI Misled You on RLHF
by macleginn on Sunday August 17, 2025 @05:47AM
(#44930289)
Everything the post says about the behaviour of OpenAI models seems to be based on pure speculation.
[ Parent
| on HN ]
Yeah, in my opinion you can just skip that part and go straight to the author's description of failing to train their own model at first and what they ended up changing to make it work: https://aerial-toothpaste-34a.notion.site/How-OpenAI-Misled-...