SKIL has a robust model storage, serving, and import system for supporting major deep learning libraries. SKIL can be used for end-to-end training, configuration, and deployment of models or alternatively you can import models into SKIL.
There are three main sources for models to be imported and hosted on SKIL:
TensorFlow models can be in the format of the Keras model file format, along with Caffe models.
These three major options give enterprise organizations the flexibility to allow multiple teams to choose their own tools while converging around a common hosting platform for deep learning and machine learning.