dstack 0.8: Local Hub, Hugging Face integration, and more

dstack, an OSS tool for running ML workflows, recently launched Hub - a new extension tailored for teams. With dstack's latest update, Hub has been made the primary method for using the tool. As a result, whether you want to run ML workflows locally or in the cloud, you now have access to all of Hub's features out of the box, including a dashboard, secure management of cloud credentials, job queue handling capabilities, and more.

Another noticeable change in the new release is the capability to run dstack Hub on Hugging Face in one click. Now, if you want to manage your workflows centrally and securely and do not want to host or deploy Hub yourself, you can do so via Hugging Face with one click. All you have to do is duplicate the dstackai/dstack-template Space template and configure your projects.

You got it right! MLOps doesn’t have to be a complicated process. It can be easy and inexpensive.

If you're interested, check the release notes on GitHub for more details.