Jordi Bolibar
Jordi Bolibar
Before finishing documentation, verify that docstrings follow the right style from SciML as inspiration.
We need to think a smart way to design an interface to specify which parameters or subparts of the equation need to be targeted for either parameter inversions or functional...
Follow the migration to Lux.jl from #151, we should remove legacy code from Flux.jl related to the `DataLoader`. Something similar exists with MLutils.jl.
We should investigate the use of [Kolmogorov-Arnold Networks](https://arxiv.org/abs/2404.19756) to train UDEs. This could improve training time, and most importanty, increase interpretability of the learnt function. This should be soon available...