It's exciting to see a novel approach to applying NNets to Tabular ML. Definitely need to call out the 1000 row limitation. It will be interesting to see if thi

It's exciting to see a novel approach to applying NNets to Tabular ML. Definitely need to call out the 1000 row limitation. It will be interesting to see if this approach stands the test of time. Other algorithms (SAINT cough, cough) made big claims, but AFAIK no one actually uses them. It's still a "XGBoost is all you need" world in tabular ML (unless you've discovered AutoGluon).