# Encoding

When feeding data into a neural network, we will need to represent out inputs (or features) in a more machine-friendly way. There are many ways to do this.

# One-hot encoding

This transforms our input set (e.g. all characters a-z) into an array that is 26-characters long. All values will be zero, except for the character we are feeding in, which will be a one. This is because the neural network doesn't give a shit about ascii or human-readable characters, it's purely operating on the underlying neurons and their inputs. See PyTorch for more info.