In this exercise we'll continue constructing the connective tissue between our data set and the yet-to-be-built neural network. In an image autoencoder, the number of input and output elements is the same as the number pixels in the image. Each pixel in the input image gets represented by one input node and each element in the output layer gets reflected by a pixel value in the output image.

It makes sense to start building this in the top-level script that we'll use to run the autoencoder. When it's operational, we can invoke it at the command line by typing python3 run_framework().

Coding challenge

• Create a run_framework.py script.
• Import the training and validation data set generators.
• Find the number of nodes that will be in the input and output (first and final) layers of the neural network.

My solution

Here's the run_framework.py code I wrote for this one.