Tabular Deep Learning

Why trees still win, attention over columns, prior-fitted networks and the benchmarks that decide the argument.