arxiv:2510.04849
Vasily Konovalov
Vasily
AI & ML interests
None yet
Recent Activity
authored
a paper
1 day ago
When Models Lie, We Learn: Multilingual Span-Level Hallucination
Detection with PsiloQA
upvoted
a
paper
2 days ago
The Rogue Scalpel: Activation Steering Compromises LLM Safety
upvoted
a
paper
2 days ago
OrtSAE: Orthogonal Sparse Autoencoders Uncover Atomic Features
Organizations
