Watch your LLM live with Visdom: Release v0.3.1

Visdom is back with a major new release.

Over the past years, contributors from the FOSSASIA community and beyond have worked on hundreds of changes across the project. The result is a much more capable Visdom with new ways to track, compare and understand machine-learning experiments, a significantly expanded visualization toolkit, new framework integrations, improved image and embedding workflows, better exports, stronger testing and major improvements across the server and frontend.

In this session, Vedansh will show what is new in Visdom and demonstrate some of the most useful features directly: experiment tracking, experiment search and comparison, hyperparameter analysis, integrations with PyTorch, Lightning, scikit-learn, Keras, XGBoost and Optuna, new visualization types, improved image workflows and many smaller improvements throughout the platform.

The new release is also the result of a growing open-source contributor community. We will look at how the project has developed, what contributors have built together and how regular collaboration is helping Visdom move forward again.

And there is more coming. We are working with developers and researchers on the next set of improvements, expanding support for modern AI workflows and developing **Visdom Dev**, a hosted platform that will make it possible to use Visdom online without operating your own server.

Visdom 0.3 is available now on GitHub and PyPI. Join the session to see what has changed, try the new…