Run a reproducible benchmark.

Launch a measured MuJoCo evaluation with explicit seeds and policies. Every run is stored in your workspace.

Ready.