An AI rewrote another AI's brain — no training. MIT had GPT-6 Astra edit Qwen's raw weights directly (just 0.016%), and its math score jumped 7.9% → 48.1%. Authors' own results, not yet replicated.▶️ https://youtu.be/qnNRWkpktKgIs weight-editing the future of fixing models — or a shortcut we don't understand?#AI #AgileFactor
An AI rewrote another AI's brain — no training. MIT had GPT-6 Astra edit Qwen's raw weights directly (just 0.016%), and ...
