ZenML adds MLFlow serving

Our friends over at ZenML have announced a new release this week. Evey ZenML release is exciting, and this is no exception. Version 0.6.2 brings improved MLFLow integration, and a better-looking CLI.

MLFlow serving

MLFLow is a pretty popular MLOPs tool that does experiment tracking, model training, as well as model serving. Already, any ZenML pipeline run could be logged to MLFLow as an experiment.

Now, you can serve the models that come out at the end of your pipelines through MLFLow as well. But there's more to this than just MLFlow:

"This is the foundation for the obvious next useful step: non-local deployments using tools like KServe and BentoML."

In other words, there will be better support for remote deployments to various serving frameworks in the future.

A new look

The command line interface has been given a new look. The release notes describe the old interface as "functional but not delightful". So they made it pretty.

First impressions count with developer tools, even if developers pretend not to mind about these things, so this is great news for adoption of ZenML.

You can read the full release notes here: https://blog.zenml.io/zero-six-two-release/