Fetching the paper…

ACTIVE-o3: Empowering MLLMs with Active Perception via Pure Reinforcement Learning · Around