Thomas Fel

Latest

• Universal Sparse Autoencoders: Interpretable Cross-Model Concept Alignment