On the Versatility of Sparse Autoencoders for In-Context Learning.

Ikhyun Cho, Gaeul Kwon, Julia Hockenmaier

Browse the full EMNLP paper archive.