Fetching the paper…

RL from Physical Feedback: Aligning Large Motion Models with Humanoid Control · Around