Summary

In this chapter, we went through the process of creating two different TensorFlow models: one using the high-level library of estimators, and the other using Keras to build a custom neural network. During this process, we also demonstrated how SageMaker can seamlessly handle the training and serving of TensorFlow models.

In the next chapter, Image Classification and Detection with SageMaker, we will explore how to use deep learning out of the box on AWS to detect and recognize images.