Re:Astra can read and write weights to improve Qwen
by versteegen on Monday November 02, 2026 @06:45PM
(#50345890)
This is actually not new, I believe GPT 5.2 already had remarkable similar capabilities. I wrote a very elaborate private benchmark involving hand-authoring highly optimised transformer circuitry, and I was shocked that GPT 5.2 could do it. Makes sense that it was specifically trained on building and debugging transformers.
[ Parent
| on HN ]
There are no comments on this comment yet.