This package contains helpers and examples for using GraphPipe with tensorflow. It contains a new plug-in operation for tensorflow that makes a call to a GraphPipe remote model from within a local tensorflow graph. The new operation is called remote_op and communicates with the remote model using libcurl and the GraphPipe protocol.
Additionaly, a new keras layer is included based on the remote operation. This allows you to include a layer in a keras model that makes a remote call.
Finally, various examples are included of serving tensorflow models in python.
For production, a more performant server like
graphpipe-tf
is
recommended, but the python server is useful for experimentation.
Building manually requires a few libraries to be installed, but the Makefile will happily run a build for you in a docker container.
make build
See build_linux.sh
for the additional headers besides libcurl that you will
need to build the C library. (From tensorflow and flatbuffers)
If you've successfully built the C library, to build installation packages:
python setup.py bdist_wheel
Note that these are not manylinux wheels and depend on libcurl being installed