original • Grasp Any Region: Towards Precise, Contextual Pixel Understanding for Multimodal LLMs

Rights note: Believed public-domain in the United States and Philippines. Report takedown concerns via the compliance support channel.