News for hackers. Stuff that matters.
Re:ChatGPT recreated with GPT-3.5? by karfly on Wednesday January 18, 2023 @01:56PM (#34431192)
IMO all RLHF stuff is mainly about aligning model not to reply with offensive and inappropriate answers, but NOT about making model's answers better.
Re:ChatGPT recreated with GPT-3.5? by xg15 on Wednesday January 18, 2023 @02:06PM (#34431360)
Ah, that makes sense. Thanks!