News for hackers. Stuff that matters.

A comment on OpenAI illegally barred staff from airing safety risks, whistleblowers say. Parent comment

Re:OpenAI illegally barred staff from airing safety risks, w... by josefx on Tuesday July 16, 2024 @08:28AM (#40975894)
It might be great for everyone if AI research had a reason to look into non brute force training methods.
[ Parent | on HN ]
Re:OpenAI illegally barred staff from airing safety risks, w... by free_bip on Tuesday July 16, 2024 @11:08AM (#40977304)
LLM training isn't even remotely brute force[1], even with tons of clever tricks to reduce the amount of compute necessary it's still fundamentally just a really hard problem.

[1]: https://youtu.be/MD2fYip6QsQ

[ Parent | on HN ]