Picture for WenBo Li

WenBo Li

Discovering Millions of Interpretable Features with Sparse Autoencoders

Add code
Jun 25, 2026
Viaarxiv icon