arXiv 论文提出两级集成黑盒方法,为 LLM 拆分偶然与认知不确定性
Black-Box Uncertainty Quantification for Large Language Models via Ensemble-of-Ensembles
用黑盒 API 就能把模型的不确定性拆成"提示有歧义"和"模型真不会"两部分,之前只有白盒方法做得到。在 5 个 QA 基准、4 个 LLM 上验证,还能扩展到 claim 级事实核查。做 RAG 或幻觉检测的工程团队值得看。
arXiv 论文(编号 2611.18453)提出一种只需黑盒采样访问的两级集成方法,用于估计大语言模型的不确定性,并能区分偶然不确定性(AU)与认知不确定性(EU)。内层通过对固定提示做随机解码构成集成,外层通过保义语义扰动提示构成集成,在连续嵌入空间中用全协方差定律做精确的组内/组间分解。
论文给出三项理论结果:嵌套蒙特卡洛的有限样本偏差恒等式及闭式偏差校正的认知估计器;连接扰动总体 AU/EU 与贝叶斯目标的贝叶斯桥梁;以及证明仅依赖一致性的黑盒方法会漏掉某类失败模式的分离结果。在 5 个短问答基准和 4 个指令微调 LLM(跨两个家族、两个规模)上,该估计器可与强白-box 基线竞争,并在相同生成预算下于错误排序上持平领先黑盒方法。
受控配对干预显示 AU 与 EU 分别对应提示歧义与模型知识缺口两类可行动的失败来源;在 NQ 长答案子集上的初步实验表明该框架无需重训练即可扩展到 claim 级事实性评估。
Black-Box Uncertainty Quantification for Large Language Models via Ensemble-of-Ensembles
Reliable use of large language models (LLMs) requires uncertainty estimates that separate ambiguity intrinsic to a prompt from genuine knowledge gaps. Bayesian and deep-ensemble methods deliver this aleatoric/epistemic decomposition in principle, but are computationally prohibitive at LLM scale and require white-box access. Existing black-box methods, in contrast, summarize to a single scalar that cannot make the distinction. We introduce a two-level ensemble that bridges this gap. Stochastic decoding of a fixed prompt forms the inner ensemble; meaning-preserving semantic perturbations of the prompt form the outer ensemble. Variability is measured in a continuous embedding space, and the law of total covariance gives an exact within-/between-perturbation decomposition to estimate aleatoric and epistemic uncertainty (AU/EU); it requires only black-box sampling access. On top of this estimator, we contribute three theoretical results that ground the framework: (i) a finite-sample bias identity for nested Monte Carlo with a closed-form bias-corrected epistemic estimator; (ii) a Bayesian bridge bounding the gap between perturbation-population AU/EU and Bayesian targets; and (iii) a separation result identifying a class of failure modes that any consistency-only black-box method provably misses while perturbation-based EU detects. Across five short QA benchmarks and four instruction-tuned LLMs spanning two families and two scales, our estimator is competitive with strong white-box baselines, and at a matched generation budget it matches leading black-box methods on error ranking while decomposing AU and EU; controlled paired interventions show that AU and EU respond selectively to distinct, actionable sources of failure: prompt ambiguity and model knowledge gaps; and a long-form pilot on the NQ long-answer split shows the framework extends to claim-level factuality without retraining.
This report is categorized under AITOP's "Papers" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include uncertainty quantification, LLM, hallucination detection, aleatoric.
Source attribution: curated from arXiv cs.LG (@Wang Ma, Debarun Bhattacharjya, Junkyu Lee, Nhan H Pham, Harsha Kokel, Qiang Ji).