Poster · Sydney
Graph-Regularized Sparse Autoencoders for LLM Safety Steering
- When
- Wed 9 Dec 2026, 10:00–13:00 AEDT
- Where
- Hall 1-4, ICC Sydney
Exhibition Centre, Level 1 - Session
- Sydney Poster Session 3
- Authors
- Jehyeok Yeon, Federico Cinus, Yifan Wu, Luca Luceri
- Organisations
- Max Planck Institute for Intelligent Systems, ELLIS Institute, Intesa Sanpaolo S.p.A, University of Southern California
- QRS classification
- Language models, Graph learning, Safety, fairness and privacy