In the last exercise we rescaled our inputs so that they would be friendly to our neural network. In this exercise we'll do the inverse. Once the autoencoder passes an image through its layers, we may want to save it or compare it to the original. We need to restore its original pixel range.

Coding challenge

• In framework.py write a denormalize() function. This will be the inverse of normalize(), transforming outputs from [-.5, 5] back to the original range.

My solution

Here is all the code we've written up to this point.