Run TAO training on your Kubernetes GPUs.
Recent revisions are kept. Restoring one creates a new revision.
