Hi, I’m Stefan! I try to understand neural networks & LLMs by analysing their internals (“mechanistic interpretability”). I work at the non-profit organization Apollo Research where I develop new mechanistic interpretability tools, and research how interpretability can enhance model evaluations for AI safety.
Contact
The best way to get in touch with me are email (firstname.lastname@gmail.com, stefan@heimersheim.eu) or DMs on LessWrong.
Legal
Responsible for the content on this site is
Stefan Heimersheim
25 Holywell Row
EC2A 4XE
London
External link disclaimer: This website may include links to third-party websites for your convenience. I do not control or endorse the content, services, or views provided by these external sites. Use of such links is at your own discretion, and I am not responsible for the privacy practices or terms of those websites.