How modern AI labs handle asynchronous reinforcement learning at scale

Original: Luke is one of the best people when it comes to RL infra, definitely worth reading!

Source: x.com ↗