logoalt Hacker News

simonw • today at 5:21 PM • 1 reply • view on HN

No, that was this one - https://rikiyat.github.io/blog/weight-space-programming/ - different authors, different approach.


Replies

vlovich123 • today at 5:36 PM

IIUC the dominant mechanism behind the math improvement remains fixing reasoning loops same as in the previous paper and probably why they picked Qwen3 here. It’s unclear this approach actually scales in any way.