Pause and Reflect: Conformal Aggregation for Chain-of-Thought Reasoning 暂停与反思:思维链推理的共形聚合 🧭 Reasoning 用加权分数聚合与共形风险控制,在有限样本保证下控制自信错误率并提升选择性准确率 Reasoning | 2026-05-13 | stat.ML | score 7/10 CoTConformal PredictionReasoningLLMUncertaintyAggregation