In this chapter, we have learned about the high-level abstractions available in TF 2.0 for model building, training, saving, and loading. Diving deep into the Keras API, we learned about how to build models by combining layers using the Sequential and functional APIs. We have also learned about how to leverage the high-level abstractions of the Keras API for training models. The chapter also looked at the intricacies of loading and saving models in various configurations and modes. We have learned about different methods of saving models, architectures, and weights, and this chapter presented an in-depth explanation of each approach and described when you should pick one over the other.
Putting together all the concepts discussed, the chapter outlined an end-to-end example program to build and train models using the Keras Sequential API. It also provided a brief overview...