Julien Herzen
Julien Herzen
See: https://github.com/unit8co/darts/issues/135#issuecomment-709979882
One possibility would be: * `u8darts` --> core (i.e., without pytorch) * `u8darts[all]` --> all (i.e., with pytorch) Shall we drop `u8darts[torch]` ?
See e.g.: https://www.lancaster.ac.uk/pg/waller/pdfs/Intermittent_Demand_Forecasting.pdf
This should be quite trivial
Currently `save_model()` saves a complete pickle of `self` (using `torch.save()`), and in addition it also saves a Pytorch Lightning checkpoint. This is because the torch-save does not save correctly the...
* Example * User guide section