I have few machines, including a workstation with powerfull GPU.
But usually, i use laptops for daily work, not powerful, but easy to take along with me.
While, i still need workstation for some heavy computing tasks/experiments, It is not so convient to switch back and forth between laptops and workstation.
and, i am tired of connecting it using remote desktop.
I found the jupyter notebook server is a good way to leverage the powerful workstation, from laptops.
- Build images and start services:
docker-compose up
- while build images, it may fail to update and install neccessary components,
RUN apt-get update && apt-get -y install gcc && pip install --upgrade setuptools && pip install -r requirements.txt
, for the sake of GFW. The solution is build with proxy arguments, check the stackoverflow link moby/moby#24697 (comment) . - the proxy port can be found in the windows proxy settings.
- Shadowsocks is easy to use in this case. default port is 1080.
- while build images, it may fail to update and install neccessary components,
- open
http://localhost
orhttp://aimachine
to open the jupyternotebook, the default password ispassword
, can be changed by updating filesrv/anaconda/conf/jupyter_notebook_config.json