gnoto is a compute service that provides GPU-powered Jupyter notebooks (link). It is well suited for debugging, visualizations, and lightweight compute tasks.
Each student can have access to a single Jupyter notebook with 1 V100 GPU with ~10 GB VRAM. The GPU for all gnoto sessions is the same.
We have already created gnoto accounts for students who are enrolled in the course. Please reach out to the teaching staff in case you face any difficulty in using your account.
Follow these quick steps:
- Go to https://gnoto.epfl.ch/.
- Click Login with Switch edu-ID and sign in.
- Wait while gnoto allocates resources to your session.
- You will land in the gnoto workspace (Jupyter-like interface).
| 1. Entry page | 2. Switch edu-ID login |
|---|---|
![]() |
![]() |
| 3. Resource allocation | 4. Gnoto workspace |
|---|---|
![]() |
![]() |
Use a custom environment when you need extra packages beyond the default gnoto setup.
my_venvs_create cs503_env
my_venvs_activate cs503_envAfter activation, install the packages you need with pip.
pip install matplotlib torchList all environments:
my_venvs_listRemove an environment you no longer need:
my_venvs_remove cs503_envA new virtual environment is available in terminal sessions only. To use it in Jupyter notebooks, register it as a kernel.
my_venvs_activate cs503_envmy_kernels_create cs503_env "CS503 Environment"Open your notebook and select the kernel from the top-right kernel selector.
If the kernel does not appear immediately, stop the notebook from Running Terminals and Kernels, reopen it, and select the new kernel.
-
Will my data be retained after I re-login into the gnoto server?
Yes. Everything in the
/home/directory will remain saved. -
How do I copy/download data into the server?
On your own computer, you can use
rsync[link] (recommended) orscp[link]. Note: these links point to SCITAS documentation, butrsyncandscpare general-purpose data transfer tools that work with any remote server, including gnoto.




