Active-O3: Empowering Multimodal Large Language Models with Active Perception via GRPO

LRM
Re-assign community