We need to first log into Github of course, if you haven't done so already run git
config --global credential.helper cache and then enter your details. Note that the
password is your access token. You can always use another authentication if you like.
I prefer to use the github command tool gh but we cannot install it here.
Once that is done, go to your home directory and clone these repositories
In your home directory, do a mkdir automl_data . This should run automatically when
Now just load Python and it's required modules so we can do the rest of the setup. You
can do this using cd /home/user/OpenML-auto-automl && ./snellius_env_loader.sh . If you
already built a conda environment, feel free to skip recreating it.
you execute it but just in case :P
Now we need to build the Singularity (It is a docker alternative. Snellius only supports
open source software so we use it) images for all the frameworks we want to run.
Since all of this is on the server, I prefer using nvim as my editor of choice.
But if you are using VSCode or some such, feel free to just open it however you want.
Of course, add whatever other frameworks you want to add.
If this does not work by any chance, you can also use cd
/home/user/OpenML-auto-automl && nohup ./build_images.sh user singularity >
build_log.txt 2>&1 &
This will take A LONG TIME to run and so the logs are stored in build_log.txt
Note that NOTHING else will work until these are built.
Alright! Now that all of this is done (hopefully) you should be good to go.