Exercise

Hyperparameter tuning for a DNN model

Hyperparameter tuning is important when attempting to create the best model for your research question. Using the tfruns package, flags can be used to iterate over several options of hyperparameter values and is a helpful way to determine the best values for each hyperparameter in a model.

In this exercise, you'll add hyperparameter tuning to the TFEStimators' DNN model you created for the Banknote Authentication dataset.

Instructions

100 XP

Call the correct function to start tuning your run.
Define the dropout flags as 0.2, 0.3 and 0.4.
Call the runs dataframe.