If you have gone through the iOS apps in the previous three chapters, you'd probably prefer using the manually-built TensorFlow iOS library to the TensorFlow experimental pod, as with the manual library method, you have more control on what TensorFlow operations can be added to make your models happy, and this better control is also one of the reasons we decided to focus on TensorFlow Mobile instead of TensorFlow Lite in Chapter 1, Getting Started with Mobile TensorFlow.
So although you can try to use the TensorFlow pod by the time you read the book to see whether the pod has been updated to support all the ops used in the model, we'll from now on always use the manually-built ...
