[BUG] TFT model load failed from saved model.

[heury]

Describe the bug

I get following error when I try to load TFT model from file.

ValueError: The values of the hyper parameter categorical_embedding_sizes should be identical between the instantiated model (None) and the loaded checkpoint ({}). Please adjust the model accordingly.

To Reproduce
I just save trained model and load it from file.
And I get following error when I try to load model from file.

ValueError: The values of the hyper parameter categorical_embedding_sizes should be identical between the instantiated model (None) and the loaded checkpoint ({}). Please adjust the model accordingly.

This error is gone when I add following parameter to the model.

categorical_embedding_sizes={}

Expected behavior

System (please complete the following information):

• Python version: 3.10
• darts version: 0.25

Additional context
Add any other context about the problem here.

[dennisbader]

Hi @heury and thanks pointing this out. It's indeed a bug and we'll fix it soon.