We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math.
It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training.
We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵
Blog:
Science
GPT-6 Astra rewrote Qwen weights to boost math 6x without training
Rikiya Takehi and collaborators had GPT-6 Astra directly edit the weights of Qwen3.5-0.8B, fixing its reasoning loops and improving math up to 6x and instruction following 3x without gradients. They call the method Weight-Space Programming.
42Klikes
5.2Kreposts
27Ksaves
4.5Mviews
- Made with
- GPT-6 Astra, Qwen3.5-0.8B
- Role
- PhD student, MIT CSAIL
- Posted
