Learning data ingestion and prediction orchestration with tensorflow from this notebook on github. It is using an InteractiveContext on the ipython environment. But obviously a production system does not run ipython. When I try to run the code in REPL python shell, I get
WARNING:absl:Method "run" is a no-op when invoked outside of IPython.
So how to run the orchestration outside a jupyter notebook, as a simple python script. Also, what other modification necessary to run the pipeline in real production environment?