Monitoring machine learning model results live from Jupyter notebooks

Tracking and saving your model results just got that much easier with

Cecelia Shao
Nov 8, 2018 · 3 min read

For many data scientists, Jupyter notebooks have become the tool of choice. Its ability to combine software code, computational output, explanatory text, and multimedia into a single document has helped countless users easily create tutorials, iterate more quickly, and showcase their work externally.

A recent Nature article cites a Github analysis that counted “more than 2.5 million public Jupyter notebooks in September 2018, up from 200,000 or so in 2015.”

The larger Project Jupyter ecosystem extends beyond the notebook — Jupyter’s newest release called JupyterLab extends the notebook framework with features such as file browsers, chat functionality, and text editors. Companies have also released tools based on notebooks where kernels reside on the cloud — most prominently, Google with their Colaboratory project.

At, we agree that Jupyter notebooks and tools will continue to play a key role within the data science community — which is why we’re very excited to release full Jupyter support with! improves the notebook experience through:

  • Instant Feedback — track your model’s results instantly and in real time
  • Low Overhead — eliminate the need to write manual code to create complex plots
  • Collaboration — share your model results easily with teammates or collaborate through teams
  • Rich Visualizations — compare results across model iterations with visualizations like bar, line, and parallel coordinates charts.

Comet is doing for machine learning what GitHub did for software. We allow data science teams to automatically track their datasets, code changes, experimentation history and production models creating efficiency, transparency, and reproducibility.

Try a sample notebook with a Keras MNIST model

Code running from Jupyter notebooks, Jupyter Lab, or any other Jupyter-based tool (such as Google Colab) now has our full support, just like your scripts!

Feel free to experiment with this sample Jupyter notebook below where we train a simple Keras model on the MNIST dataset in order to conduct image classification (i.e. classify a handwritten 7 as a 7). This notebook demonstrates how to use Comet to automatically track the model code, results, and more.

You can also play with the full public Comet project here with a browser view.

With, you can finally:

  • have a tight feedback loop between EDA and modeling
  • view real-time model results plotted as nice visualizations
  • easily collaborate with other data scientists

Try with your own Jupyter Notebook

Sign up for your free account at and explore Comet with your own Jupyter-based tools! We’re very excited about this new full integration and look forward to hearing your feedback on Github or Slack.

Found this article useful? Follow us ( on Medium and check out some other relevant articles below! Please 👏 this article to share it!

Build better models faster

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store