Submitted by Chirag Agarwal 9 Towards Understanding the Robustness of Sparse Autoencoders Aikyam Lab 2 2