Patch-as-Decodable-Token: Towards Unified Multi-Modal Vision Tasks in MLLMs Paper • 2510.01954 • Published 16 days ago • 11
