AnalysisAI ModelsNovember 2, 2026

MIT CSAIL: GPT-6 Astra edits Qwen weights, lifts MATH-500 7.9% to 48.1%

Read original source →venturebeat.com

In "Sparks of Weight-Space Programming," four MIT CSAIL researchers report GPT-6 Astra natively edited 0.016% of Qwen3.5-0.8B's weights, fixing reasoning loops and raising MATH-500 from 7.9% to 48.1% with no gradients. IFEval rose 3x; Astra also wrote small game networks from scratch that outplayed PPO-trained ones.

People · Omar Khattab, Alex Zhang

How this story unfolded

1 day · 2 reports · 12 community posts · 14 of 16 shown

  1. Nov 2
  2. Nov 3

More stories today

Open the live feed