The RL is done on problems with verifiable answers. I’m not sure how o1 slop would be at all useful in that respect.