by Luma AI · Model · Robotics
27B robot policy trained only in simulation on 40 tasks, handles new real-world tasks; only a paper so far