Introduction

In the previous chapters, you learned how to build CNN or RNN models from predefined TensorFlow modules. You have been using one of the APIs offered by TensorFlow called the sequential API. This API is a great way to start building "simple" deep learning architecture with few lines of code. But if you want to achieve higher performance, you may want to build your own custom architecture. In this case, you will need to use another API called the functional API. Researchers use functional APIs while defining their model architecture. By learning how to use it, you will be able to create custom loss functions or modules, such as a residual block from the ResNet architecture.