Skip to content
Autocompletion with Deep Learning on Jupyter Notebook
JavaScript Python Go Shell CSS Dockerfile
Branch: master
Clone or download
Latest commit d07bd8d Sep 16, 2019

TabNine for Jupyter Notebook

Read this in other languages: English, 中文


This extension for Jupyter Notebook enables the use of coding auto-completion based on Deep Learning.

Other client plugins of TabNine require starting a child process for TabNine binary and using Pipe for communication. This can’t be done with Jupyter Notebook, since child process can’t be created with JQuery and Jupyter Notebook doesn’t provide any way for adding third-part js libs to plugins.

In this repository, it is achived by developing a client plugin and a server plugin for Jupyter Notebook. The client plugin generate request info and send http request to the server plugin. The server plugin pass the request info to it’s client process (TabNine) and return the request to client plugin.


The extension consists of a pypi package that includes a javascript notebook extension, along with a python jupyter server extension. Since Jupyter 4.2, pypi is the recommended way to distribute nbextensions. The extension can be installed:

  • from the master version on the github repo (this will be always the most recent version)

  • via pip for the version hosted on pypi

From the github repo or from Pypi,

  1. install the package

    • pip3 install [--user][--upgrade]

    • or pip3 install jupyter-tabnine [--user][--upgrade]

    • or clone the repo and install

      git clone

      python3 install

  2. install the notebook extension jupyter nbextension install --py jupyter_tabnine [--user|--sys-prefix|--system]

  3. and enable notebook extension and server extension

    jupyter nbextension enable --py jupyter_tabnine [--user|--sys-prefix|--system]
    jupyter serverextension enable --py jupyter_tabnine [--user|--sys-prefix|--system]

For Jupyter versions before 4.2, the situation after step 1 is more tricky, since the --py option isn’t available, so you will have to find the location of the source files manually as follows (instructions adapted from @jcb91’s jupyter_highlight_selected_word). Execute

python -c "import os.path as p; from jupyter_tabnine import __file__ as f, _jupyter_nbextension_paths as n; print(p.normpath(p.join(p.dirname(f), n()[0]['src'])))"

Then, issue

jupyter nbextension install <output source directory>
jupyter nbextension enable jupyter_tabnine/main
jupyter serverextension enable <output source directory>

where <output source directory> is the output of the first python command.


  • Jupyter TabNine will be active after being installed. Sometimes, you may want to show the Jupyter original complete temporally, then click shift + space.

    show original complete demo

  • Remote auto-completion server is also supported. You may want this to speed up the completion request handing. Or maybe your company want to deploy a compeltion server cluster that services everyone. Read following to learn how to deploy remote server.


Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

Please make sure to update tests as appropriate.


GNU General Public License v3.0

Remote Completion Server Deployment

You can’t perform that action at this time.