From token probabilities to calibrated confidence: An empirical study of mathematical question answering

• URL: http://arxiv.org/abs/2608.07827v1
• Date: Sat, 08 Aug 2026 00:00:19 GMT
• Title: From token probabilities to calibrated confidence: An empirical study of mathematical question answering
• Abstract summary: Prior work has shown that token probabilities are often overconfident.<n>We compare single-pass estimators, which reuse token probabilities from the original generation, with multi-pass estimators.<n>We find that aggregating token probabilities over the full sequence captures small but consistent differences between correct and incorrect generations.
• Score: 1.0032961794537367
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Confidence estimation for large language models (LLMs) aims to estimate the probability that a generated answer is correct, while calibration aligns these estimates with empirical accuracy. Prior work has shown that token probabilities are often overconfident, we investigate whether these readily available signals can nevertheless provide well-calibrated confidence estimation for mathematical question answering. We compare single-pass estimators, which reuse token probabilities from the original generation, with multi-pass estimators, which obtain additional confidence signals through verification or stochastic forward passes. While individual token probabilities can be highly saturated, we find that aggregating token probabilities over the full sequence captures small but consistent differences between correct and incorrect generations, yielding more informative confidence estimates. Multi-pass methods can yield calibrated confidence estimates. We study two such approaches: self-verification through re-prompting, including a lower-cost in-situ variant, and Monte Carlo Dropout, which derives confidence from variation across stochastic forward passes. We further evaluate two post-hoc calibration methods, Platt scaling and isotonic regression, both of which substantially reduce in-domain calibration error. However, their data efficiency varies with dataset difficulty, and the calibration mappings often transfer asymmetrically across datasets and models.

Related papers

• DirEAG: Dirichlet Evidence Aggregation for Calibrating Verbalized Confidence in Mathematical Reasoning [2.867517731896504]
  Black-box verbalized confidence is difficult to calibrate in large language models.<n>We propose DirEAG, a Dirichlet Evidence Aggregation method that converts each elicited answer-confidence observation into calibrated soft evidence.
  arXiv  Detail & Related papers  (2026-08-21T03:48:25Z)
• Post-Calibration Reliability Reranking of Relevance Decisions via Label-wise Monotone Projection [35.67041829703622]
  Web search, product search, and question-answering retrieval systems often assign a relevance label and confidence score to each query-candidate pair.<n>Post-hoc calibration is needed because misaligned confidence can make systems over-trust wrong predictions or unnecessarily defer correct ones.<n>We address this gap with Label-wise Monotone Reliability Projection (MRP), which learns label-wise monotone functions that map calibrated confidence to correctness reliability.
  arXiv  Detail & Related papers  (2026-08-11T02:51:35Z)
• Discovery of Hidden Miscalibration Regimes [52.452902154360565]
  Models may be systematically overconfident on some kinds of inputs and underconfident on others.<n>We define the corresponding miscalibration field and propose a diagnostic framework for estimating it.<n>Our approach learns a calibration-aware representation of the input space and estimates signed local miscalibration by kernel smoothing in the learned geometry.
  arXiv  Detail & Related papers  (2026-05-13T13:07:50Z)
• Geometric Calibration and Neutral Zones for Uncertainty-Aware Multi-Class Classification [0.0]
  This work bridges information geometry and statistical learning, offering formal guarantees for uncertainty-aware classification in applications requiring rigorous validation.<n> Empirical validation on Adeno-Associated Virus classification demonstrates that the two-stage framework captures 72.5% of errors while deferring 34.5% of samples, reducing automated decision error rates from 16.8% to 6.9%.
  arXiv  Detail & Related papers  (2025-11-26T01:29:49Z)
• Provably Reliable Conformal Prediction Sets in the Presence of Data Poisoning [53.42244686183879]
  Conformal prediction provides model-agnostic and distribution-free uncertainty quantification.<n>Yet, conformal prediction is not reliable under poisoning attacks where adversaries manipulate both training and calibration data.<n>We propose reliable prediction sets (RPS): the first efficient method for constructing conformal prediction sets with provable reliability guarantees under poisoning.
  arXiv  Detail & Related papers  (2024-10-13T15:37:11Z)
• Revisiting Confidence Estimation: Towards Reliable Failure Prediction [53.79160907725975]
  We find a general, widely existing but actually-neglected phenomenon that most confidence estimation methods are harmful for detecting misclassification errors. We propose to enlarge the confidence gap by finding flat minima, which yields state-of-the-art failure prediction performance.
  arXiv  Detail & Related papers  (2024-03-05T11:44:14Z)
• Binary Classification with Confidence Difference [100.08818204756093]
  This paper delves into a novel weakly supervised binary classification problem called confidence-difference (ConfDiff) classification. We propose a risk-consistent approach to tackle this problem and show that the estimation error bound the optimal convergence rate. We also introduce a risk correction approach to mitigate overfitting problems, whose consistency and convergence rate are also proven.
  arXiv  Detail & Related papers  (2023-10-09T11:44:50Z)
• On double-descent in uncertainty quantification in overparametrized models [24.073221004661427]
  Uncertainty quantification is a central challenge in reliable and trustworthy machine learning. We show a trade-off between classification accuracy and calibration, unveiling a double descent like behavior in the calibration curve of optimally regularized estimators. This is in contrast with the empirical Bayes method, which we show to be well calibrated in our setting despite the higher generalization error and overparametrization.
  arXiv  Detail & Related papers  (2022-10-23T16:01:08Z)
• Don't Just Blame Over-parametrization for Over-confidence: Theoretical Analysis of Calibration in Binary Classification [58.03725169462616]
  We show theoretically that over-parametrization is not the only reason for over-confidence. We prove that logistic regression is inherently over-confident, in the realizable, under-parametrized setting. Perhaps surprisingly, we also show that over-confidence is not always the case.
  arXiv  Detail & Related papers  (2021-02-15T21:38:09Z)
• Evaluating probabilistic classifiers: Reliability diagrams and score decompositions revisited [68.8204255655161]
  We introduce the CORP approach, which generates provably statistically Consistent, Optimally binned, and Reproducible reliability diagrams in an automated way. Corpor is based on non-parametric isotonic regression and implemented via the Pool-adjacent-violators (PAV) algorithm.
  arXiv  Detail & Related papers  (2020-08-07T08:22:26Z)

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.