osc.viz.attentions.viz_vit_attns

viz_vit_attns(images, attns_dict, head_reduction='mean', adjust_residual=True)[source]