Route Sparse Autoencoder to Interpret Large Language Models.

Wei Shi, Sihang Li, Tao Liang, Mingyang Wan, Guojun Ma, Xiang Wang, Xiangnan He

Browse the full EMNLP paper archive.