Meituan team combines reinforcement learning with self-teaching for agents

Original: The meituan team is easily one of my favorites. It's hard to put my finger on it but I love their research directions and aesthetics

Source: x.com ↗