Revisiting Multimodal Positional Encoding in Vision-Language Models Paper • 2510.23095 • Published Oct 27 • 20