On the Limits of Sparse Autoe | Pangram Labs