- Clone the repo and
cd
to project root:git clone https://github.com/deepmipt/stand_squad_ru.git cd stand_squad_ru
- Run script to download and unpack model components:
./download_components.sh
- Create a virtual environment with
Python 3.6
:virtualenv env -p python3.6
- Activate the environment:
source ./env/bin/activate
- Install requirements:
pip install -r requirements.txt
- Download NLTK data:
$ python3 >>> import nltk >>> nltk.download('punkt')
- Specify model endpoint host (
api_host
) and port (api_port
) insquad_agent_config.json
- Specify
CUDA_VISIBLE_DEVICES
and virtual environment path (if necessary) inrun_ru_squad.sh
- Run model:
./run_ru_squad.sh
-
If necessary, build Base Docker image from:
-
Clone the repo and
cd
to project root:git clone https://github.com/deepmipt/stand_squad_ru.git cd stand_squad_ru
-
Build Docker image:
sudo docker build -t stand/squad_ru .
-
Run Docker image:
sudo docker run -p <host_port>:6005 --runtime=nvidia --device=/dev/nvidia<gpu_unit_int_id> -v </path/to/host/vol/map/dir>:/logs stand/squad_ru