News for hackers. Stuff that matters.

A comment on Why Sam Altman Won't Be on the Hook for OpenAI's Spending Spree. Parent comment

Re:Why Sam Altman Won't Be on the Hook for OpenAI's Spending... by locknitpicker on Saturday November 08, 2025 @11:39AM (#45857979)
> Taking that at face value, it means we would have to invest exponential resources just to get linear improvements.

Not necessarily. Approaches such as mixture of experts help lower training costs by covering domains with specialized models.

[ Parent | on HN ]
Re:Why Sam Altman Won't Be on the Hook for OpenAI's Spending... by diamond559 on Saturday November 08, 2025 @11:57AM (#45858122)
haha good one, so why haven't they done this yet? What are they waiting for? Let's see these super advanced "experts" with "specialized models"!!
[ Parent | on HN ]
Re:Why Sam Altman Won't Be on the Hook for OpenAI's Spending... by locknitpicker on Saturday November 08, 2025 @01:21PM (#45858735)
> haha good one, so why haven't they done this yet? What are they waiting for? Let's see these super advanced "experts" with "specialized models"!!

I understand it's very easy to post ignorant messages in internet forums, but the answer to your question is yes, "they have done it" and it does result in cheaper training costs. See models such as DeepSeek-MoE or Mixtral.

https://github.com/deepseek-ai/DeepSeek-MoE

https://mistral.ai/news/mixtral-of-experts

[ Parent | on HN ]
Re:Why Sam Altman Won't Be on the Hook for OpenAI's Spending... by hiddencost on Saturday November 08, 2025 @11:41AM (#45858000)
I encourage you to rethink your identity. You are way out of your depth on this, and posting nonsensical things as fact.
[ Parent | on HN ]