1. Connect to WIFI (patreo or patreo_5GHz) or VPN

2. Connect to servers (GPU: grenoble, cergy, pontoise; CPU: lille) via SSH:

Examples:

ssh edemir@cergy

3. Create a docker container with support to gpus:
docker run --gpus all -it --name=$(whoami)_DESCRICAO -v /home/users/$(whoami):/home -w /home edemirfaj/patreo_bot:gpu-py3 bash

Obs: The image above "edemirfaj/patreo_bot:gpu-py3" contains pytorch '1.4.0' and tensorflow '1.14.0' installed on python3 (3.6.8); the flags -p 6006 and -p 8888 randomly maps the common used ports for tensorboard (6006) and jupyter notebooks (8888), which can be check using "docker ps | grep $(whoami)" outside the container.