Positive values are orange, negative values are blue, zero is white.
There must be an input and output layer at minimum.
There can be any number of hidden layers, but they must be named with increasing numbers: layer0, layer1, layer2, etc.
Each layer is an array of any number of neurons.
Each neuron must have a set of weights, a bias, and an activation function.
The number of weights a neuron must have is equal to the number of neurons in the previous layer.
The output layer is defined as a single neuron.
The input layer is an array that can have any number of values. The values are strings that are used in the GLSL. You have access to a vec2 named "pos" that gives the 2d coordinate of the pixel, but can do any operations to those values like square them, add a constant, put them through sin, pow, etc.
You can share the URL to have someone see exactly what you are seeing.