Submitted by Chirag Agarwal 10 Towards Understanding the Robustness of Sparse Autoencoders Aikyam Lab 2 2