How to deploy pytorch model using C++

Application
#1

I have trained a pytorch model. I want to use it for inference on MIPS platform running embedded OS based on Linux. I am struggling to find necessary steps in Apache TVM docs. Here is what I have done so far:

  1. Translate pytorch model to TorchScript model as explained here
  2. Cross compiled TVM runtime for my MIPS platform as explained here.

However I am unable to find any further steps.

My target platform does not run python, but only C/C++. Does the documentation contain any example / tutorial that explain how can I deploy pytorch model using C++?

#2

Perhaps refer to this document?

https://tvm.apache.org/docs/how_to/deploy/cpp_deploy.html

It should be possible to run inference in your language of choice without the need for Python.