通过组合一致性检测大模型推理错误，无需标签即可精准识别失败。
Operadic consistency: a label-free signal for compositional reasoning failures in LLMs
- 基于操作代数理论，检验模型直接答案与分解推理结果是否一致。
- 在4个数据集上相关性达0.86~0.94，优于现有所有置信度指标。
- 适用于开源与闭源模型，可提升选择性预测准确率，适合可靠性评估场景。
在无真实标签的情况下检测大模型推理失败，已有诸多置信度基线方法，如自一致性、语义熵和P(True)，这些方法依赖问题内采样与自评估。操作代数理论（operad theory）——用于描述通过迭代替换构建的系统——提出了互补诊断：模型对组合型问题的直接回答应与其通过显式分解推导出的答案一致。本文提出操作一致性（OC），一种逐题信号。在12个指令微调的大模型（参数量4B至671B，开源与闭源）上测试，覆盖4个多跳问答数据集，OC与准确率高度相关（皮尔逊相关系数r ∈ [0.86, 0.94]，p ≤ 0.0004），且是唯一在全部四个数据集上保持r ≥ 0.85的信号。链式思维自一致性（CoT-SC）仅在HotpotQA和DROP上表现良好（r = 0.93, 0.87），在MuSiQue和StrategyQA上降至约0.45。在单题层面，OC提供的信息超越CoT-SC与语义熵（聚类稳健性p ≤ 10⁻¹⁶），且在控制构造分解感知基线后结论仍成立（p ≤ 10⁻¹³）。在相同成本预算K=3下，使用该信号的择优预测性能优于调优后的CoT-SC基线，AUARC提升0.086~0.096，AUROC提升0.092~0.164，95%置信区间均不含零。在5个前沿思维模型上，分解由模型自身链式思维提取，相同比较在全部16个（数据集，预算，指标）组合中均显示正向点估计提升，12组95%置信区间不含零。
原文摘要 · Abstract (English)
Detecting LLM reasoning failures at inference time without ground-truth labels has motivated a wide range of confidence baselines, including self-consistency, semantic entropy, and P(True), built on within-question sampling and self-evaluation. Operad theory, the formalism for systems built by iterated substitution, suggests a complementary diagnostic: a model's direct answer to a compositional query should agree with the answer it produces by composing a stated decomposition of the same query. We instantiate this idea as operadic consistency (OC), a per-question signal. Across twelve instruction-tuned LLMs (4B to 671B parameters, open-weights and closed-source) on four multi-hop QA datasets, OC is strongly correlated with accuracy on every dataset (Pearson $r \in [0.86, 0.94]$, all $p \leq 0.0004$), and is the only signal we evaluate with $r \geq 0.85$ uniformly across all four datasets. Chain-of-thought self-consistency (CoT-SC; Wang et al., 2023) matches OC on HotpotQA and DROP ($r = 0.93, 0.87$) but drops to $r \approx 0.45$ on MuSiQue and StrategyQA. At the per-question level, OC contributes information beyond CoT-SC and semantic entropy on every dataset (cluster-robust $p \leq 10^{-16}$ for the OC coefficient), and the conclusion is robust to additionally controlling for constructed decomposition-aware baselines ($p \leq 10^{-13}$). The same signal yields selective-prediction improvements (accuracy at fixed coverage) over a tuned CoT-SC baseline at the equal-cost $K = 3$ budget (AUARC lifts of +0.086 to +0.096 and AUROC lifts of +0.092 to +0.164; 95% CIs exclude zero on every cell). On five frontier thinking models, where the decomposition is extracted from the model's own chain of thought, the same equal-cost comparison gives positive selective-prediction point-estimate lift on all 16 (dataset, budget, metric) cells tested, with 95% CIs excluding zero on 12 of the 16.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。