ClearML Agent is the ClearML MLOps agent which in conjunction with the ClearML Hosted Service (or self-hosted ClearML Server) provides full AI cluster automation. It supports replicating a Task’s (experiments and other workflows) original environment on target GPU machines, and executing that Task with tracking and resource monitoring; job scheduling; pre-configuring Docker containers for Task execution; ClearML services for long-lasting jobs; and other use cases (see the ClearML Agent use case examples).

Any ClearML user can use a running instance of ClearML Agent.

Only someone running ClearML Agent needs to install and configure it.

To run ClearML Agent, first install and configure it on the target GPU machine. For a list of ClearML commands, see the ClearML Agent Reference.