News for hackers. Stuff that matters.

A comment on Some Thoughts on Autoregressive Models. Parent comment

Re:Some Thoughts on Autoregressive Models by energy123 on Friday March 07, 2025 @04:51AM (#43288752)
Yann LeCun's prediction was empirically refuted. He says that the longer LLMs run, the less accurate they get. OpenAI showed the opposite is true.
[ Parent | on HN ]
Re:Some Thoughts on Autoregressive Models by mrfox321 on Friday March 07, 2025 @05:51AM (#43289112)
They didn't show this, they just increased the length where accuracy breaks down.
[ Parent | on HN ]
Re:Some Thoughts on Autoregressive Models by energy123 on Friday March 07, 2025 @06:00AM (#43289166)
Explain? OpenAI showed the new scaling law in December 2024 that performance keeps increasing proportional to ln(N reasoning tokens)
[ Parent | on HN ]
Re:Some Thoughts on Autoregressive Models by mentalgear on Friday March 07, 2025 @07:17AM (#43289555)
link?
[ Parent | on HN ]