Nice. I had never heard of TVM before. Hopefully I can find a tutorial in their docs to get a quick feel for using it.

I have experimented with saving Keras models and loading and running in Racket Scheme - very different I know, but I am interested in deploying DL models.