this post is so fucking stupid to the point of being embarrassing. wtf was the author doing? Tune both and see how much better they are vs each other? But they just fucking change some parameters without hyperparameter-tuning training and show TabNet was better.