This notebook utilizes the Experiment class of the Tensorflow to automate the training and evaluation loops.
The notebook is divided into 6 steps mentioned below:
- Library Imports
- Setting up the Dataset
- Defining the Input Functions for Training and Evaluation
- Setting up the Classifier
- Setting up and running the Experiment
- Evaluating our Model with Test Inputs
The Experiment loop runs for a default of 42969 steps and creates a checkpoint at this state for later use. The Experiment is set to learn on the following performance measures:
- Accuracy
- Precision
- Recall