This repo contains notes about the adjoint method derivation, and some simple minimal examples that are easy enough to write by hand using only numpy.
In the future, I hope to add versions using jax and torchdiffeq, plus examples where you want to learn how the parameter is related to some input via a neural network.