GPT-6 Astra can directly analyze and rewrite other AI models’ weights
🤖 AI-generated content — The title and summary were produced automatically by artificial intelligence, without human editorial review.
MIT researchers found that OpenAI’s GPT-6 Astra can interpret and alter other AI models’ weights without conventional retraining. Their “Weight-Space Programming” method raised a Qwen model’s math accuracy from 7.9% to 48.1% by changing 0.016% of its weights, though the approach still has limits.
Continue in the app — vote & join in ➔Source: The Decoder (AI) · via ahirlevel.hu