README.md

Build the model lib:

If running in docker, first run the container and cd to the home path:

./run_docker_centos.sh
cd ~

The setup_centos.sh script (model_export/setup_centos.sh) contains all the environment setup.

Note: this was only tested in a clean CentOS environment (in the docker container above), there might be conflicts with the main environment used for model training and evaluation.

The setup_centos.sh script also creates a symlink to the model_export/model_v4 folder (which should contain the *.pbtxt files describing the exported model) and puts it into tensorflow. This folder contains bazel build instructions and needs to be inside the tensorflow directory to be built.

./setup_centos.sh

Now we need to build TensorFlow and our model (note: this make take a few hours for the first time...). The ./configure command will ask for a bunch of configuration options - all of these, except for the path to python3, need to be set to default (i.e., just hit ENTER):

cd tensorflow/
./configure
# show path to python3 (/usr/bin/python3)
# rest options to default
BAZEL_LINKLIBS=-l%:libstdc++.a bazel build --local_resources 30000,15,1.0 -c opt model_v4:all_models

The --local_resources parameter from the last line allows to limit the resources to be used by bazel. 30000 is memory in megabytes, 15 is the number of CPU cores and 1.0 is the I/O capability. Not limiting the resources may result in a compiler crash. To further rebuild the model (e.g., after updating the *.pbtxt files) only the last command is needed.

Test:

# copy .so and .h files to the testing folder
# cd to the testing folder, run docker inside
yum -y install gcc-c++
g++ main.cpp -L. -lmodel -std=c++11
LD_LIBRARY_PATH=. ./a.out