Value of model_path in upload_model() function of move_model.py incorrect for ensemble models

[kh296]

The upload_model() function of move_model.py uses (line 73):

model_path=str(model_path / "final_model"

This is fine for models trained in a single run, but for models trained with multiple cross-validation splits the model path is:

model_path=str(model_path / "final_ensemble_model"

[ant0nsc]

Thanks for flagging this. @JonathanTripp , can you take a look?