“We have a phrase we call the Moore's Law of RL environments, which basically means, like, every six to eight months, we've noticed that the complexity of the environment… roughly doubles.”

“They're applying these models to the RL environments themselves… hack this environment… fixing those… lower rates of cheating downstream”

“Seed RL… was about utilizing the TPUs both for inference and training in the best possible way while keeping the environments as flexible as you wanted.”

“He filed a motion to dismiss on double jeopardy grounds… the law says that there is manifest necessity here because… extended deliberations… judge gave renewed instructions.”

“… very good RL controllers that are probably trained from motion imitation, and then fine-tuned with good RL that suits the robot's body… for the specific task of running really fast.”