AnalysisAI ModelsNovember 3, 2026

LLMs can natively read and write neural network weights

Read original source →rikiyat.github.io

GPT-6 Astra edited 0.016% of Qwen3.5-0.8B's weights without gradients, lifting MATH-500 from 7.9% to 48.1% and tripling IFEval. It also wrote small game-playing networks from scratch that outplay PPO-trained nets given 100M steps.

1 source

More stories today

Open the live feed