The MapR Data Science Refinery container includes a FUSE-based MapR POSIX Client, optimized for containers, that allows deep learning libraries to read and write data directly to MapR-FS.
So, when you run TensorFlow, the compute occurs on the host where the container resides, but each container has full access to the persistent storage provided by the MapR Converged Data Platform. When you kill the container off, the data remains.
Complete Steps to deploy this can be found here.
Rachel Silver is the Product Management Lead for Machine Learning & AI @ MapR Data Technologies.