Excuse me, if this question seems to be redundant, but searching through the documentation, github repository + issues and the discuss page could not answer my questions.
I want to run a small network, which has been trained in keras on an RPi4 (standalone, C++). I was able to convert the keras model and even ran the AutoTVM with RPC on the device and saved the resulting lib.tar, params and graph.json files.
But now I am very confused on how to proceed with that: I looked into
- howto_deploy and
- bundle_deploy
but could not figure out, which way to go and how to adapt these examples for my needs. Have I overlooked a step here? I extracted the lib0.o from the .tar file, but was not able to load it in cpp_deploy.cc