neural ordinary differential equations (nodes)

These models use ordinary differential equations to model the transformation of a neural network's hidden states over continuous space, providing a different approach to depth in networks.