Using projects, you can now organize related runs by grouping them together. For example, if you're generating a personalized product recommendation model, you can create a project called 'Purchase Predictions' to group relevant runs. Now, all future runs you or your collaborators create to solve this problem can also be added here for better organization and collaboration.
Within projects, you’ll find some handy features. You can now select specific metrics you’d like to use to track progress and create a key metric graph that visualizes how you’re moving toward your goals. In addition, columns on tables are customizable — add, remove, or reorder them and include any metric or hyperparameter as its own column. Improved filters also allow you to filter by any metric or hyperparameter.
Multiple Orgs for Enterprise Users
To allow for more flexibility with organization needs, enterprise users now have the ability to host multiple orgs, enabling internal machine and data separation. This allows each org to run on their own cluster and house their own resources.
Service Mesh for Model Serving for Enterprise Users
We’ve increased security in model serving, providing full encryption of all traffic between nodes in your model serving cluster.
Markdown Support in ‘Notes’ Fields
A small, but useful update, you can now write in markdown and LaTeX in ‘Notes’ fields, and we also support code syntax highlighting in markdown. Try it out!
Dependencies from Virtual Environments
Now, you can use your currently active virtualenv, conda env, pipenv, or poetry environment for dependency management.
Coming Soon, Experiments!
Look out for the experiments feature, which will make it possible to do a deep dive on select runs within a project.