JupyterLab

Overview

Teaching: 0 min
Exercises: 0 min

Questions

• What is JupyterLab?

Objectives

• Learn: number of GPU instances, selecting a Docker image and GPU memory

Users can deploy one or more private JupyterLab applications.

To encourage fair sharing these applications are time limited.

Selecting a number of GPU instances

Number of GPU instances

• The AF cluster has four NVIDIA A100 GPUs.

• GPU partitioned into -> seven GPU instances.

• AF cluster can have up to 28 GPU instances running in parallel.

• You can select anywhere from 0 to 7 GPU instances as resources for the notebook.

Selecting Docker image

2 image options

• 1: full anaconda (ivukotic/ml_platform:conda)

• 2: NVidia GPU and ROOT support (ivukotic/ml_platform:latest)

  • This one has ML packages (Tensorflow, Keras, ScikitLearn,…) preinstalled, check /ML_platform_tests tutorial

For software additions and upgrades please contact ivukotic@uchicago.edu

Selecting GPU memory

Select 40,836 MB for an entire A100 GPU. Select 4864 MB for a MIG instance.

You can learn more about in its JupyterLab documentation link

Key Points

• Check JupyterLab documentation