The main purpose is to provide OpenAI's Gym with Tensorflow and GUI.
- User control (unprivileged)
- VNC (tightvncserver)
- GUI (xfce4).
- CUDA 9.0 (with CUDNN 7)
- Python: Miniconda
- Build from the source
- Make sure all sh files are executable! Otherwise do the similar after deploy (in the container).
find . -type f -name "*.sh" -exec chmod +x {} \;