Policies trained from data instead of hand-written controllers.
Open weights and code for Physical Intelligence's π0 robot foundation models.
