Currently model training is done via the WW_MLP3_Training_pytorch.ipynb notebook, and analysis is done in the other notebooks by reading in the artifacts it produces.
Keras training needs to be replicated in a WW_MLP3_Training_pytorch.ipynb notebook that produces similar artifacts.