ODEs are just one dimensional PDEs. The documentation of NeuralPDE has quite a few examples of this:
https://docs.sciml.ai/NeuralPDE/dev/tutorials/ode/
https://docs.sciml.ai/NeuralPDE/dev/tutorials/param_estim/
https://docs.sciml.ai/NeuralPDE/dev/manual/ode/
Note that we are changing the loss in these docs examples to make it more robust: Docs modifications to use QuadratureTraining instead of GridTraining by sdesai1287 · Pull Request #729 · SciML/NeuralPDE.jl · GitHub