Dear , 

We are happy to announce that we will be officially changing the  Trains  product name. The official announcement will be made in January, but we wanted to give you a heads-up as we will be soft-launching this change starting this week. The new name is...

< drumroll>  

:alphabet-white-c: :alphabet-white-l: :alphabet-white-e: :alphabet-white-a: :alphabet-white-r: :alphabet-yellow-m: :alphabet-yellow-l:

🚨 But wait, there’s more! 🚨

We are happy to give you a heads-up on the impending launch - together with the formal name change in January - of a  managed  service of ClearML. An online hosted version of the  Trains  ClearML you all love and cherish, with three exceptionally helpful features: 

Secure login with either Google, GitHub, or that other thing. Share experiments with an external link - for viewing by any ClearML user.  Collaborate with your colleagues with  shared  workspaces.

Initially, the launch will be for a completely free hosted managed service limited to three users on a shared workspace and 100GB of storage per workspace.

Already this coming week, we will unveil the website and the hosted service of ClearML. It will be available side-by-side with all the existing Trains packages. These packages will continue to work indefinitely.

In early January - at a date that we will share with you in advance - the Trains Git repository will change its name to ClearML. Nonetheless, your existing packages, servers, agents, etc., will continue to work indefinitely. However, all new releases going forward will be under the ClearML name. No further new releases will be made under the Trains name.

🤯 FAQ 🤯

I have explicitly integrated Trains into my code. Do I have to change anything?

Actually, not just yet. While the packages will be interchangeable in the short-term (so only search and replace),  we will be deprecating the Trains package “real soon.”

 I have deployed my own  trains-server . What does this mean to me?

At first, nothing. The ClearML package and trains-server will work together… for a while. Eventually, as more features are released (really cool stuff coming very soon) this will motivate you to install a  clearml-server  instance. The transition should be seamless and… we are here if there are any problems.

Can I connect to ClearML with my trains or trains-agent?

Most definitely. Just set up your app credentials on ClearML, then point your existing installation through its configuration files to the ClearML server.

What do I need to do after January’s launch when I want to download the new versions of  clearml  and  clearml-agent ? How do I make the transition from Trains?

Just  pip install  and start using the new package: Import  clearml  into your python code, and launch your ClearML workers using the  clearml-agent . Same old dance, new name, and adding new capabilities.

What about the existing experiments and data on my  trains-server ?

Deploying  clearml-server  will be almost, if not entirely, the same as deploying trains-server up until now. Your data is safely stored on your machine. Your new  clearml-server  installation will make use of your existing data.

Does the fact that you are launching ClearML managed service mean you will no longer support ClearML as an open-source package for self-serving?

ClearML will carry on the Trains torch as a free, open-source solution. You’ll be using the same open-source python package and agent regardless of which server you use. You can use the new free managed service with its usage constraints, contact our sales team to learn about the enterprise version and its additional capabilities and services, or carry on using the open-source server you deployed free and clear.