github link:
https://github.com/Nordeus/heroic-rl

## File Structure

ct2_train
├── Dockerfile
└── train_volume
    └── CgiVideoGameFramework
        └── VideoGameLearningFramework
            ├── requirements.txt
            ├── setup.py
            └── vglf
                ├── heroic_rl
                └── ...

## Setup

podman build -t heroic_image -f Dockerfile .
podman pod create --name heroic_net -p 6006:6006
podman run --pod heroic_net -d -v $PWD/train_volume:/root -w /root --name=heroic -it heroic_image tmux
podman exec -it heroic tmux a
cd ~
chmod -R 755 .
cd CgiVideoGameFramework/VideoGameLearningFramework/
pip install -e .
pip install -r requirements.txt
pip install joblib==0.13.2
apt install libopenmpi-dev
pip install mpi4py==3.0.3
pip install psutil==5.7.0

## Run
Open heroic server:

podman run -it --pod heroic_net -d --name="heroic_server" quay.io/nordeus/heroic-rl-server:latest

Run train code:

podman exec -it heroic tmux a
(env CUDA_VISIBLE_DEVICES=0) python run.py -f heroic.yaml (-k ppo -i 4)