Process Reward Models that Think -- https://arxiv.org/abs/2504.16828