Back to Publications
2026Information Sciences

Low-rank sparse autoencoders: Unifying efficiency and geometric regularization for large language model interpretability

Mu, Jiajia, Tan, Benying, Ren, Jie, Guo, Mingyu, Luo, Chenchen, and Bai, Ruibin

Keywords

sparse autoencoderslarge language modelsmechanistic interpretabilitygeometric regularizationcomputational efficiencyinference latency

Authors from this organization

Ruibin Bai

Ruibin Bai

Director of Lab

Computer Science and Operations Research

View on Publisher Site