git clone <repo>
cd <repo-dir>
virtualenv -p python3.5 venv # or newer
source venv/bin/activate
PyTorch is not on PyPi, so this needs to be installed separately:
pip install http://download.pytorch.org/whl/cu80/torch-0.2.0.post3-cp35-cp35m-manylinux1_x86_64.whl"
pip install .
This can be run several times if the variables needs setting multiple times (needed also on client side).
source setup.sh
For now, you have to launch the server from the repo directory in order for the server to find the trained models.
python python_src/server.py