JupyterLab Learning

Run popular AI/ML tutorials in a JupyterLab* environment on Intel® Tiber™ AI Cloud.

Tip

Click Learning in the console main menu to view Jupyter Notebooks.

Prerequisites

Overview

A JupyterLab* environment is provided for users who want to study, create, and share executable code. By default, Jupyter Notebooks are pre-loaded with popular libraries and tools, including Intel® software and Intel® hardware. See Hardware and software.

  1. From the console main menu, select Learning.

  2. Choose an option.

    • Launch Jupyter Notebook – Most common

      • Launch the default Python 3 (ipykernel)

      • Dependencies are loaded via import statements

      • Pre-existing code is included

    • Launch JupyterLab

      • Launch the default Python 3 (ipykernel)

      • Notebooks launch as bare, without pre-loaded modules or code.

Launch Jupyter Notebook

  1. Select a Jupyter Notebook from a category, such as AI with Intel® Gaudi® AI accelerator.

    Note

    This is the most common option for students and developers.

  2. Select the Launch button.

  3. Follow the instructions in the Notebook.

  4. Place your cursor in the cell to execute code.

  5. Press SHIFT+ENTER to run code and review output.

You’re all set.

Launch JupyterLab

  1. Click on the button Launch JupyterLab.

  2. At the dialog, “Your JupyterLab access is ready”, select Launch.

  3. At the prompt Select Kernel, choose base to run the Notebook.

  4. In a cell, enter your code and import modules.

  5. Place your cursor in the cell to execute code.

  6. Select SHIFT+ENTER to run code and review output.

Notebook Tips

  • To change cell behavior, select the Code pull-down menu at top. Optional: Change to another option, like Markdown for text, or Raw.

  • To change the kernel, select the pull-down menu in upper right. When Select kernel appears, select your preferred kernel. See also Jupyter kernels.

Hardware and software

Hardware

The JupyterLab environment provides access to various types of hardware, including:

  • Intel® Gaudi® 2

  • Intel® Gaudi® AI Accelerator

  • Intel® Data Center GPU Max Series

Hardware access operates in a shared model where each user accesses a portion of CPU, GPU, and memory resources. For dedicated or exclusive access, use bare metal or virtual machine instances. See Manage Instance.

Software

The JupyterLab environment includes a variety of commonly used software to help accelerate learning. The table below shows a subset of the available Intel® software installed in the JupyterLab environment.

Installed software

Software

Version

PyTorch*

2.5

Intel® Distribution for Python

2024.2.0

Intel® oneAPI Base Toolkit

2024.2.1

Intel® oneAPI HPC Toolkit

2024.2.1

Intel® Quantum SDK

1.1

Intel® oneAPI DPC++/C++ Compiler

2024.2.1

Intel® Fortran Compiler & Intel® Fortran Compiler Classic

2024.2.1

Intel® Distribution for GDB*

2024.2.1

Intel® Integrated Performance Primitives

2021.12.1

FPGA Add-on Package Compiler - Community Support

2024.2.1

Intel® Integrated Performance Primitives Cryptography

2021.12.1

Intel® oneAPI DPC++ Library

2022.6.1

Intel® oneAPI Threading Building Blocks

2021.13.1

Intel® Trace Analyzer and Collector

2022.2.0

Intel® Advisor

2024.2.1

Intel® oneAPI Data Analytics Library

2024.6.0

Intel® oneAPI Collective Communications Library

2021.13.1

Intel® oneAPI Deep Neural Network Library

2024.2.1

Intel® oneAPI Math Kernel Library

2024.2.1

Intel® VTune™ Profiler - Community Support

2024.2.1

Intel® DPC++ Compatibility Tool

2024.2.1

Intel® MPI Library

2021.13.1

Kernel options

Jupyter utilizes a kernel model to allow different configurations of software that can be selected for a given Notebook.

Each notebook has a default kernel. Automatically, a kernel runtime is set for each Notebook upon launch. However, you can select a different kernel in a Notebook after you launch it. For a comprehensive list of Jupyter kernels, visit Jupyter documentation

Available kernels

Kernel

Intel® Max Series GPU

Intel® Gaudi® 2 AI Accelerator

Base

✓

Modin

✓

Python 3

✓

✓

PyTorch

✓

PyTorch 2.5

✓

PyTorch GPU

✓

TensorFlow

✓

TensorFlow GPU

✓

Sessions

A session is an environment provided to a single user to interact with Jupyter for a set duration of time. To conserve resources, the system will terminate a session:

  • If you explicitly log out by clicking “File” on the Jupyter top navigation and selecting “Log Out.”

  • If your Jupyter session is idle for 15 minutes the system will automatically log you out of a session.

  • When the maximum 4-hour session duration is elapsed, the session is logged out.

You may create a new session at any time by following the same access instructions.

Session lifecycle

Accounts that have not used Jupyter for 20 days are considered inactive and may have all local file contents removed. Each new session will reset this rolling 20-day timer, and users who have had their environments removed can always launch a notebook again.

Updating libraries

Libraries are updated regularly to ensure you have access to the latest versions. If you require a specific library or version that does not exist, we encourage you to post on in our Community page.

Note

New libraries and tools are frequently added to the JupyterLab environment. Such dependencies are subject to change without notice.

User-installed packages

Optionally, you may install your own software via the Jupyter terminal. For example, you may do so by downloading binary files, cloning code repositories, or utilizing a package manager such as conda or pip. When executed, such code is installed by default in the ~/.local directory. Similarly, when Jupyter notebook guide users to pip install required libraries, these libraries are installed into ~/.local directory.

Troubleshooting

If you encounter issues with the JupyterLab environment, try one of the following.

Restart Kernel

In the Notebook, select the Kernel menu, and choose an action (e.g., “Restart kernel”).

Caution

Take care to save or back up the code entered in cells. Choosing some menu options may delete code depending on its state.

Logout

  1. Select File` on the top navigation bar and then select Log Out.

  2. Relaunch the Notebook by clicking the Launch.

  3. Evaluate if a relaunch corrects the issue.

Policy

The standard terms and conditions of the Intel® Tiber™ AI Cloud Service Agreement apply.

Fees JupyterLab resources are currently provided free of charge. However, you acknowledge and agree that this policy is subject to change without notice solely applies to JupyterLab resources, and not any other Intel® Tiber™ AI Cloud services, software, or professional services.

Use Restriction. As articulated in the Intel® Tiber™ AI Cloud Service Agreement, use of the JupyterLab resources for illegal or malicious activities is strictly prohibited, and Intel reserves the right to suspend or remove access at any time.