In the previous chapter, we have seen that there are two ways to define a model in TensorFlow:

• Premade estimators
• Building a custom estimator

In the following code, we will consider one additional snippet of code that would enable us to visualize the various summary operations:

Note that we only need to specify the model_dir in the premade estimator to store the various log files generated from TensorFlow operations.

TensorBoard can then be initialized by referring to the model directory, as follows:

The preceding code would result in a TensorBoard visualization that would have all the summaries built in.