Train Deformable DETR to find objects in images.
Recent revisions are kept. Restoring one creates a new revision.
