Hi,
I’d like to ask, is the material on feature columns still relevant? I am talking about the “# Classify structured data with feature columns” notebook, since the authors themselves say that should be avoided for new code in favor of using Keras’s built-in layers for preprocessing.