##article.return## Supervised Reinforcement Learning: From Expert Trajectories to Step-wise Reasoning Download Download PDF