Forums

Hi! By any chance, is it technically possible to deploy TF Serving on PythonAnywhere to serve prediction requests from remote clients?

You'd need to create a web app for that, but tensorflow doesn't work well currently in our web apps. For more details have a look at this help page.

Thank you for replying! Actually, my question is around Tensorflow Serving package, not Tensorflow itself. What it does is to import an already existing neural network and to provide predictions via REST. It is not used to create neural networks. But looks like it's not a common idea anyways.

Unless it has a WSGI interface, it will not work on PythonAnywhere.

Thanks much for clarifying!