Some extra experimental evaluation of learned-indices: https://dawn.cs.stanford.edu/2018/01/11/index-baselines/