GPT-6 Astra rewrites AI model weights directly, MIT shows
MIT researchers show GPT-6 Astra can read and rewrite other AI models' weights directly, lifting Qwen3.5-0.8B's MATH-500 score from 7.9 to 48.1 percent.
Topic
Model Editing
Topic
Topic
Topic
MIT researchers show GPT-6 Astra can read and rewrite other AI models' weights directly, lifting Qwen3.5-0.8B's MATH-500 score from 7.9 to 48.1 percent.