CRoS：打乱推理步骤顺序的因果奖励，提升模型准确率与CoT忠实度
CRoS: Causality Reward on Scrambled CoTs Improves Accuracy and Reasoning Faithfulness
一篇训练方法的论文：打乱推理步骤顺序来算因果奖励，Gemma-3-4B 在没见过的数据集上准确率从 43.8% 提到 46.8%，忠实度也翻倍。
论文提出 CRoS，一种新的 CoT 因果性度量方法，通过打乱推理步骤的顺序而不改动内容来扰动 CoT，从而避免删除步骤或注入错误带来的信息变化干扰。在 Reasoning Gym 上用 Gemma-3-4B 做实验，CRoS 在五个未见过的逻辑与游戏数据集上取得 46.8% 准确率，超过最强基线的 43.8%。CoT 质量方面，CRoS 在 CoT-Pass@K 从 K=1 到 128 的区间内平均得 43.4%，基线为 41.4%。CoT 忠实度上，CRoS 的 hint acknowledgment 得分为 34.0%，明显高于基线的 20.2%。
CRoS: Causality Reward on Scrambled CoTs Improves Accuracy and Reasoning Faithfulness
Reinforcement learning with verifiable rewards (RLVR) has become the standard method for training language models to reason. However, RLVR typically relies solely on outcome correctness, which does not guarantee that the CoT causally contributes to the final answer. Without this connection, outcome rewards may fail to reinforce high-quality reasoning. Existing causality metrics assess CoT causality by perturbing the CoT, such as deleting reasoning steps or injecting errors, and measuring changes in the final answer. However, these perturbations inevitably alter information in each step. Consequently, high causality scores may reflect sensitivity to information changes introduced by perturbation rather than true causality. Such confounding effects further induce undesirable reasoning behaviors when these metrics are used as training rewards, including generating less informative CoTs and reducing exploration and self-correction. To address this problem, we propose CRoS (Causality Reward on Scrambled CoTs), a novel CoT causality metric that perturbs the order of reasoning steps without altering content, thereby preserving information within each step. On Reasoning Gym with Gemma-3-4B, CRoS outperforms existing causality and outcome rewards in out-of-domain accuracy, CoT quality, and CoT faithfulness. Specifically, CRoS achieves 46.8% accuracy on five unseen logic and games datasets, outperforming the strongest baseline (43.8%). It also achieves better CoT quality, consistently outperforming baselines in CoT-Pass@K from K=1 to 128, averaging 43.4% vs. 41.4%. For CoT faithfulness, CRoS achieves a hint acknowledgment score of 34.0% vs. 20.2%.