Summary

This chapter demonstrated how to build and utilize custom TensorFlow components. You learned how to design and implement custom loss functions, layers, and residual blocks. Using the TensorFlow functional API or model subclassing allows you to build more complex deep learning models that may be a better fit for your projects.

In the next and final chapter, you will explore and build generative models that can learn patterns and relationships within data, and use those relationships to generate new, unique data.