Storing fitting model parameters

[armarion]

Is there a simple way to modify the code in DifferentialMLTF2.ipynb to store the fitted neural net and load the stored parameters back into the regressor for later prediction?