“We're in this paradigm of offline RL. You go and you procure high-quality datasets from data vendor companies. You go and hill climb those specifically”

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

“We put a ton of time into not just model architecture, but also the data pipelines around it, how we eval the product, and having a really good RL loop around post-training and mid-training the model”

“When you apply this Goldilocks algorithm of RL, you can actually push the models to a lot better performance”

“… 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.”