News for hackers. Stuff that matters.

A comment on GPT-5.5. Parent comment

Re:GPT-5.5 by minimaxir on Thursday April 23, 2026 @06:02PM (#47882708)
Golden Gate Claude was two years ago and it's surprising there hasn't been as much research into targeted activations since.
[ Parent | on HN ]
Re:GPT-5.5 by landl0rd on Thursday April 23, 2026 @09:07PM (#47884308)
There’s been some, but naive activation steering makes models dumber pretty reliably and training an SAE is a pretty heavy lift.
[ Parent | on HN ]