Andrei Sakhovskii
Boolalg
AI & ML interests
None yet
Recent Activity
upvoted
a
paper
about 5 hours ago
OrtSAE: Orthogonal Sparse Autoencoders Uncover Atomic Features
upvoted
a
paper
about 5 hours ago
The Rogue Scalpel: Activation Steering Compromises LLM Safety
Organizations
None yet