LLMs can edit model weights without gradient descent
LLMs are starting to optimize neural networks directly, letting researchers change a model for a natural-language goal without writing a training loop or using gradients. In one test, GPT-6 Astra edited just 0.016% of Qwen3.5-0.8B’s weights to reduce reasoning loops, raising MATH-500 from 7.9% to 48.1% and tripling IFEval. The same approach also identified what small game networks had learned from their raw weights.