GPT-6 Astra fixed Small Broken Qwen Model in One shot!

13.6K views · 181 likes · Nov 3, 2026

Did OpenAI really teach a small AI model math by fixing its weights? Not quite. A team at MIT CSAIL (Rikiya Takehi, Jayden Teoh, Alex Zhang and Omar Khattab) used OpenAI's GPT-6 Astra to read and edit the raw weights of Qwen3.5-0.8B, with no training and no gradients. Astra found…

0 comments

Sign in to comment.

No comments yet. Be the first.

More videos