{"ID":6023317,"CreatedAt":"2026-07-08T01:00:23.257252134Z","UpdatedAt":"2026-07-10T00:55:22.603132029Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2607.05721","arxiv_id":"2607.05721","title":"SpanUQ: Span-Level Uncertainty Quantification for Large Language Model Generation","abstract":"Uncertainty estimation is essential not only for the trustworthy deployment of large language models (LLMs) but also as a foundation for self-refinement in LLM generation. However, existing approaches operate at suboptimal granularities: token-level scores lack semantic coherence, while sequence-level scores fail to localize errors. We formalize Span-Level Uncertainty Estimation (SLUE), a new task that targets the natural granularity for uncertainty: semantically coherent text spans, each conveying a single assessable unit of meaning. To address this task, we introduce SPANUQ, a lightweight probe that distills the uncertainty knowledge from expensive multi-sample inference into a single forward pass over LLM hidden states. SPANUQ employs a DETR-style span decoder to simultaneously detect spans and estimate their uncertainty via a Mixture of Beta distribution, trained with a principled combination of Beta NLL regression and contrastive ranking objectives. We construct SPANUQ-BENCH, the first span-level uncertainty benchmark comprising 20K prompts, 293K annotated spans, and continuous soft labels derived from multi-sample claim verification. Experiments on five LLM backbones show that SPANUQ consistently achieves the best span-level uncertainty quality, outperforming the strongest probe baseline and all sampling-based methods while being 10-20x faster. Its DETR-based span detector attains 0.910 F1, surpassing the best heuristic by 39.4%, enabling precise error localization that sequence-level methods cannot provide. The framework generalizes across five LLMs spanning two model families.","short_abstract":"Uncertainty estimation is essential not only for the trustworthy deployment of large language models (LLMs) but also as a foundation for self-refinement in LLM generation. However, existing approaches operate at suboptimal granularities: token-level scores lack semantic coherence, while sequence-level scores fail to lo...","url_abs":"https://arxiv.org/abs/2607.05721","url_pdf":"https://arxiv.org/pdf/2607.05721v1","authors":"[\"Yimeng Zhang\",\"Yingying Zhuang\",\"Ziyi Wang\",\"Yuxuan Lu\",\"Pei Chen\",\"Aman Gupta\",\"Zhe Su\",\"Ming Tan\",\"Zhilin Zhang\",\"Qun Liu\",\"Manikandarajan Ramanathan\",\"Rajashekar Maragoud\",\"Edward Vul\",\"Jing Huang\",\"Dakuo Wang\"]","published":"2026-07-07T01:09:46Z","proceeding":"cs.CL","tasks":"[\"cs.CL\"]","methods":"[\"Large Language Model\",\"Language Model\"]","has_code":false}