In this tutorial, we want to integrate Keras models with Tensorflow Estimators. Estimator is a powerful feature that brings flexibility to run an experiment and helps developers to avoid boilerplate code for training/testing a model. Check this interesting video for an introduction to Estimator. Keras provides an abstract api for developing neural networks and helps developer for fast prototyping. By reading this tutorial, you will be able to Keras models in Tensorflow ecosystem.
Check the jupyter notebook for details.