The example for monotonicity and faithfulness given is on scikit model. What about keras NN? The keras based predict_proba is a 1D list while the sciket based predict_proba is a 2D list. Any tutorial to implement monotonicity and faithfulness on NN using keras dense layers?