Deriving Decoder-Free Sparse Autoencoders from First Principles

Open in new window