TopoNets: High-Performing Vision and Language Models with Brain-Like Topography