Using TensorFlow Serving

In this section, we will show you how to set up your RNN model to predict spam or ham text messages on TensorFlow. We will first illustrate how to save a model in a protobuf format, and will then load the model into a local server, listening on port 9000 for input.

Getting ready

We start this section by encouraging reader to read through the official documentation and the short tutorials on the TensorFlow Serving site available at https://www.tensorflow.org/serving/serving_basic.

For this example, we will reuse most of the RNN code we used in the on Predicting Spam with RNNs recipe in Chapter 9, Recurrent Neural Networks. We will alter our model saving code to save a protobuf model in the correct folder...