Version conflicts

[Blackandwhite23]

I tried to use your sample code, but I didn't get it running. It seems I am always running into a version conflict.
I tried to pip install the project, but it couldn't find the tensorflow versions below 2. Then I tried to install a lower version of python (3.6), then I could find some tensorflow versions, but not 1.13.0rc1. So I installed 1.13.2 but afterwards I got error messages from torch ("No module named 'torch._C'"), so I guess I am also using the wrong torch version.
Is it possible to know which versions of which module you used for your test and describe that anywhere?

[Blackandwhite23]

If I use python=3.7.16 and tensorflow==1.13.1, then I get the message
"TypeError: Descriptors cannot not be created directly.
If this call came from a _pb2.py file, your generated code is out of date and must be regenerated with protoc >= 3.19.0."

If I use
"pip install protobuf==3.20.*"
afterwards, it seems I only get warnings:
" FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_qint8 = np.dtype([("qint8", np.int8, 1)])"