Accessing GPUs through Dedicated Deployment's Jupyter Notebook

I’ve been trying to use the Dedicated Deployment’s dev-gpu through Jupyter Notebook. But after starting the Dedicated deployment and launching the notebook, the following code inside a .ipynb file returns False:

import torch
torch.cuda.is_available()

Also, when I run the command nvidia-smi in the Jupyter Notebook terminal, I get the following error:

bash: nvidia-smi: command not found

Do I need to install any additional packages or make some configuration changes to enable the dedicated deployment’s GPU usage on the Jupyter Notebook?

I also ran the deployment with machine type selected as GPU.

Hi @Dikshant_Shah!
Are you using a GPU compatible docker image?

For further reference, here are the docs on how to deploy a dedicated deployment.

Oh no. I was trying to access the Dedicated GPU through the Jupyter Lab Notebook. Seems like the referenced docs doesn’t have anything about this. This is the feature:

This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.