Workstation
You can use the lab workstation to train models or do any other compute-intensive tasks. The workstation user name is labmember
. Once you’ve logged in, create your own personal folder inside the home directory (e.g., ~/jhennig
).
Setting up SSH access
To ssh into the lab workstation, you will need to do a few things:
-
If you have not already done so before, create an ssh key on your personal computer, with
ssh-keygen -t rsa -b 4096
-
Copy the ssh public key on your personal computer (
~/.ssh/id_rsa.pub
) to the workstation’s~/.ssh/authorized_keys
for thelabmember
user. -
On your personal computer, create or append to the file
~/.ssh/config
:
Host henniglab
HostName WORKSTATION_IP
User labmember
Port WORKSTATION_PORT
where WORKSTATION_IP
and WORKSTATION_PORT
are numbers I or another lab member can tell you.
You should now be able to log in from your personal computer with ssh henniglab
.
Using the workstation’s GPU in a Jupyter notebook
TBD
General guidelines
When installing python packages, only install to a conda environment (or any other python environment manager), not the default python installation. For example, the following will create a new conda environment:
conda create --name valuernn python=3.9 pytorch matplotlib numpy scipy scikit-learn
conda activate valuernn
Once you have activated your environment, you can always install additional packages with:
conda install thepackageiforgot