Benchmarking Uncertainty Calibration in Large Language Model Long-Form Question Answering

• URL: http://arxiv.org/abs/2602.00279v1
• Date: Fri, 30 Jan 2026 20:02:34 GMT
• Title: Benchmarking Uncertainty Calibration in Large Language Model Long-Form Question Answering
• Authors: Philip Müller, Nicholas Popovič, Michael Färber, Peter Steinbach,
• Abstract summary: Large Language Models (LLMs) are commonly used in Question Answering (QA) settings.<n>Existing UQ approaches remain weakly validated in scientific QA.<n>We introduce the first large-scale benchmark for evaluating UQ metrics in reasoning-demanding QA.
• Score: 7.1559850008795385
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Large Language Models (LLMs) are commonly used in Question Answering (QA) settings, increasingly in the natural sciences if not science at large. Reliable Uncertainty Quantification (UQ) is critical for the trustworthy uptake of generated answers. Existing UQ approaches remain weakly validated in scientific QA, a domain relying on fact-retrieval and reasoning capabilities. We introduce the first large-scale benchmark for evaluating UQ metrics in reasoning-demanding QA studying calibration of UQ methods, providing an extensible open-source framework to reproducibly assess calibration. Our study spans up to 20 large language models of base, instruction-tuned and reasoning variants. Our analysis covers seven scientific QA datasets, including both multiple-choice and arithmetic question answering tasks, using prompting to emulate an open question answering setting. We evaluate and compare methods representative of prominent approaches on a total of 685,000 long-form responses, spanning different reasoning complexities representative of domain-specific tasks. At the token level, we find that instruction tuning induces strong probability mass polarization, reducing the reliability of token-level confidences as estimates of uncertainty. Models further fine-tuned for reasoning are exposed to the same effect, but the reasoning process appears to mitigate it depending on the provider. At the sequence level, we show that verbalized approaches are systematically biased and poorly correlated with correctness, while answer frequency (consistency across samples) yields the most reliable calibration. In the wake of our analysis, we study and report the misleading effect of relying exclusively on ECE as a sole measure for judging performance of UQ methods on benchmark datasets. Our findings expose critical limitations of current UQ methods for LLMs and standard practices in benchmarking thereof.

Related papers

• On Calibration of Large Language Models: From Response To Capability [66.59139960234326]
  Large language models (LLMs) are widely deployed as general-purpose problem solvers.<n>We introduce capability calibration, which targets the model's expected accuracy on a query.<n>Our results demonstrate that capability-calibrated confidence improves pass@$k$ prediction and inference budget allocation.
  arXiv  Detail & Related papers  (2026-02-14T01:07:45Z)
• Boosting Process-Correct CoT Reasoning by Modeling Solvability of Multiple-Choice QA [10.122669382758122]
  We show that when questions are effectively unsolvable for a model, spurious chains of thought (CoTs) are more likely to appear.<n>We adapt outcome-supervised reward models and reinforcement learning with group-relative advantage to incorporate solvability into their objectives.<n>Our results highlight solvability as a key factor for reducing hallucinations and increasing reliability in CoT reasoning.
  arXiv  Detail & Related papers  (2025-09-30T08:34:16Z)
• UQ: Assessing Language Models on Unsolved Questions [149.46593270027697]
  We introduce UQ, a testbed of 500 challenging, diverse questions sourced from Stack Exchange.<n>UQ is difficult and realistic by construction: unsolved questions are often hard and naturally arise when humans seek answers.<n>The top model passes UQ-validation on only 15% of questions, and preliminary human verification has already identified correct answers.
  arXiv  Detail & Related papers  (2025-08-25T01:07:59Z)
• Revisiting Uncertainty Quantification Evaluation in Language Models: Spurious Interactions with Response Length Bias Results [10.551985027162576]
  Evaluations often use metrics like AUROC to assess how well UQ methods correlate with task correctness functions.<n>We show that mutual biases--when both UQ methods and correctness functions are biased by the same factors--systematically distort evaluation.<n>We identify LM-as-a-judge methods as the least length-biased, offering a promising path for a fairer UQ evaluation.
  arXiv  Detail & Related papers  (2025-04-18T13:13:42Z)
• Right Answer, Wrong Score: Uncovering the Inconsistencies of LLM Evaluation in Multiple-Choice Question Answering [78.89231943329885]
  Multiple-Choice Question Answering (MCQA) is widely used to evaluate Large Language Models (LLMs)<n>We show that multiple factors can significantly impact the reported performance of LLMs.<n>We analyze whether existing answer extraction methods are aligned with human judgment.
  arXiv  Detail & Related papers  (2025-03-19T08:45:03Z)
• Reliable and Efficient Amortized Model-based Evaluation [57.6469531082784]
  The average score across a wide range of benchmarks provides a signal that helps guide the use of language models in practice.<n>A popular attempt to lower the cost is to compute the average score on a subset of the benchmark.<n>This approach often renders an unreliable measure of LM performance because the average score is often confounded with the difficulty of the questions in the benchmark subset.<n>We train a model that predicts question difficulty from its content, enabling a reliable measurement at a fraction of the cost.
  arXiv  Detail & Related papers  (2025-03-17T16:15:02Z)
• CoT-UQ: Improving Response-wise Uncertainty Quantification in LLMs with Chain-of-Thought [10.166370877826486]
  Large language models (LLMs) excel in many tasks but struggle to accurately quantify uncertainty in their generated responses.<n>Existing uncertainty quantification (UQ) methods for LLMs are primarily prompt-wise rather than response-wise.<n>We propose CoT-UQ, a response-wise UQ framework that integrates LLMs' inherent reasoning capabilities through Chain-of-Thought.
  arXiv  Detail & Related papers  (2025-02-24T14:48:06Z)
• Legitimate ground-truth-free metrics for deep uncertainty classification scoring [3.9599054392856483]
  The use of Uncertainty Quantification (UQ) methods in production remains limited.<n>This limitation is exacerbated by the challenge of validating UQ methods in absence of UQ ground truth.<n>This paper investigates such metrics and proves that they are theoretically well-behaved and actually tied to some uncertainty ground truth.
  arXiv  Detail & Related papers  (2024-10-30T14:14:32Z)
• Benchmarking Uncertainty Quantification Methods for Large Language Models with LM-Polygraph [83.90988015005934]
  Uncertainty quantification is a key element of machine learning applications.<n>We introduce a novel benchmark that implements a collection of state-of-the-art UQ baselines.<n>We conduct a large-scale empirical investigation of UQ and normalization techniques across eleven tasks, identifying the most effective approaches.
  arXiv  Detail & Related papers  (2024-06-21T20:06:31Z)
• Word-Sequence Entropy: Towards Uncertainty Estimation in Free-Form Medical Question Answering Applications and Beyond [52.246494389096654]
  This paper introduces Word-Sequence Entropy (WSE), a method that calibrates uncertainty at both the word and sequence levels. We compare WSE with six baseline methods on five free-form medical QA datasets, utilizing seven popular large language models (LLMs)
  arXiv  Detail & Related papers  (2024-02-22T03:46:08Z)
• Towards Clear Expectations for Uncertainty Estimation [64.20262246029286]
  Uncertainty Quantification (UQ) is crucial to achieve trustworthy Machine Learning (ML) Most UQ methods suffer from disparate and inconsistent evaluation protocols. This opinion paper offers a new perspective by specifying those requirements through five downstream tasks.
  arXiv  Detail & Related papers  (2022-07-27T07:50:57Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.