TopoNets: High performing vision and language models with brain-like topography