High-Level Libraries for TensorFlow

There are several high-level libraries and interfaces (API) for TensorFlow that allow us to build and train models easily and with less amount of code such as TF Learn, TF Slim, Sonnet, PrettyTensor, Keras and recently released TensorFlow Estimators.

We will cover the following high-level libraries in this chapter while dedicating the next chapter to Keras:

• TF Estimator - previously TF Learn
• TF Slim
• TFLearn
• PrettyTensor
• Sonnet

We shall provide examples of building the models for MNIST dataset using all of the five libraries. Do not worry about understanding the details of the models yet as we cover the details of models from chapter 4 onwards.

You can follow the code examples in this chapter with the Jupyter Notebook ch-02_TF_High_Level_Libraries included in the code bundle. Try modifying the examples in the notebook to experiment and play...