Skip to main content

Overview

JupyterLite is a responsive python notebook environment we provide for your learners' best experience.

Should you use JupyterLite?​

For python-notebook labs, we strongly recommend you use JupyterLite for your learners' best experience.

However, sometimes JupyterLite may not be an option for you - we lay out these cases in the next sections.

For more information on why JupyterLite is preferred over JupyterLab see JupyterLite vs JupyterLab.

Supported Libraries​

Preinstalled libraries​

JupyterLite comes with a set of preinstalled Python packages. You can find a list of these packages here. The preinstalled packages include some popular Python libraries such as numpy, pandas and sklearn. In order to use these packages, you simply have to import them after an optional %pip install:

### OPTIONAL ###
%pip install numpy pandas scikit-learn
################

import numpy as np
import pandas as pd
import sklearn

Note that the aforementioned list of preinstalled packages relates to the latest version of JupyterLite that can be deployed. However, the JupyterLite version that is actually deployed by Skills Network is not necessarily the latest version, so the actual list of preinstalled libraries as well as the preinstalled libraries' versions may differ somewhat from that list.

Libraries from PyPI​

Pure python packages with wheels on PyPI can be installed. Pure python packages are those that have files on PyPI with the pattern *py3-none-any.whl. An example of such a package is seaborn: if you go to https://pypi.org/project/seaborn/#files you will notice a file with a pattern matching *py3-none-any.whl. In order to install pure python packages remember to pin the version you wish to install, otherwise your lab may break when the package is updated on PyPI:

%pip install seaborn==0.13.2

Unsupported or Partially Supported Libraries​

Packages that are not preinstalled and not pure python packages cannot be installed. Moreover, because access to the internet is restricted in the JupyterLite environment, packages that rely on an established internet connection may not work even if they can be installed; for these packages the workaround at Downloading Data might work for you and you may not need to use the non-working package at all.

The following popular python libraries are not supported or only partially supported in JupyterLite:

If your lab requires an unsupported package and a suitable alternative or workaround does not exist, you should use JupyterLab instead.

Additional Caveats​

Even if your lab uses libraries supported by JupyterLite, there are specific scenarios where you may want to consider using JupyterLab:

  • Your lab has highly cpu-intensive code.
    • For example, sklearn.manifold.TSNE.fit_transform was found to be impractical in JupyterLite.
  • Large datasets (>100MB+) are being used.
    • Loading large datasets may be inconvenient for your learners on JupyterLite - they will load faster with JupyterLab.
  • You need a specific version of a Python package that differs from the one that is preinstalled AND the preinstalled package is not a pure python package.
    • JupyterLite does not allow you to upgrade or downgrade preinstalled packages unless they are pure python packages. For instance, %pip install numpy==1.26.3 will fail, but %pip install tqdm==4.67.1 will succeed even if another version of tqdm is preinstalled.
  • You need to use two or more versions of a Python package in one notebook
    • JupyterLite does not allow you to upgrade or downgrade packages once they have been %pip installed or imported. For instance, %pip install seaborn==0.13.1 followed by %pip install seaborn==0.13.2 will fail at the second install.

Using JupyterLite​

Installing Packages​

Installing packages in JupyterLab is usually done with a package manager such as pip, conda or mamba like:

!pip install numpy pandas
!conda install -c conda-forge numpy pandas
!mamba install numpy pandas

In JupyterLite you simply use the %pip magic command:

%pip install numpy pandas

For pure python packages we strongly recommend pinning the version you wish to install:

%pip install optuna==4.2.0

For additional information about supported packages and caveats read the Supported Libraries, Unsupported or Partially Supported Libraries and Additional Caveats sections.

Downloading Data​

You may be accustomed to loading data the following way:

import pandas as pd

URL = 'https://www.url.to/my/dataset.csv'

df = pd.read_csv(URL)

In JupyterLite, you must use the following method instead:

import pandas as pd
import skillsnetwork

URL = 'https://www.url.to/my/dataset.csv'

await skillsnetwork.download_dataset(URL)
df = pd.read_csv('dataset.csv')

Relying on online sources carries the risk of labs becoming non-functional if those sources are moved or become unavailable. Therefore, we strongly recommend transferring these data sources to the Skills Network File Library before using them in your labs. To upload files to the Skills Network File Library, follow the steps provided in Skills Network File Library - Access File Library within Jupyterlite or Jupyterlab.

Best practices​

Follow these best practices when creating JupyterLite labs:

  • Do not ask your learners to restart the kernel after %pip install
    • Restarting the kernel is common after installing packages in JupyterLab. However, restarting the kernel after installing packages in JupyterLite will make those packages unavailable in JupyterLite
  • Pin pure python package versions when installing from PyPI.
    • For instance, instead of using %pip install optuna, use %pip install optuna==4.2.0

Convert from JupyterLab to JupyterLite​

You can easily convert from JupyterLab to JupyterLite by following Convert Labs from JupyterLab to JupyterLite (and Vice Versa) - Convert JupyterLab to JupyterLite.

Convert from JupyterLite to JupyterLab​

You can easily convert from JupyterLite to JupyterLab by following Convert Labs from JupyterLab to JupyterLite (and Vice Versa) - Convert JupyterLite to JupyterLab.