$ ssh foo@bar -L 5900:localhost:5900 -L 6006:localhost:6006 -L 8888:localhost:8888
$ docker run --gpus all -it -p 5900:5900 -p 6006:6006 -p 8888:8888 --name dl_remote naruya/dl_remote
% vnc # start x11vnc
% jl # jupyter lab
% tb # tensorboard
python -c "import torch; print(torch.cuda.is_available())"
% apt install x11-apps mesa-utils
% vnc # start x11vnc
% xeyes # cpu rendering
% glxgears # gpu rendering