Hoppa till innehåll
VibekollenBETAVibekollen
TelegramTechmeme

Researchers say LLMs can be good at natively reading and writing neural network weights; GPT-6 Astra edited Qwen3.5-0.8B's weights to make it better at math

rikiyat.github.io via Techmeme

TL;DR … If you were asked to handcraft a neural network that plays Go, what would you do?

If you were handed raw neural network weights …

Kort telegram från Techmemes öppna flöde, på källans språk. Hela originalartikeln har inte lästs.

Mer från Techmeme