The Model class implements a model that stores a neural network and input/output channels, and links these. To execute a model on different backends, you can download the model object into an object of the ModelExecutor class and call the start() model execution function.