Annotation-Efficient Universal Honesty Alignment

• URL: http://arxiv.org/abs/2510.17509v1
• Date: Mon, 20 Oct 2025 13:05:22 GMT
• Title: Annotation-Efficient Universal Honesty Alignment
• Authors: Shiyu Ni, Keping Bi, Jiafeng Guo, Minghao Tang, Jingtong Wu, Zengxin Han, Xueqi Cheng,
• Abstract summary: Existing methods either rely on training-free confidence estimation or training-based calibration with correctness annotations.<n>Elicitation-Then-Calibration (EliCal) is a two-stage framework that first elicits internal confidence using inexpensive self-consistency supervision, then calibrates this confidence with a small set of correctness annotations.<n>EliCal achieves near-optimal alignment with only 1k correctness annotations (0.18% of full supervision) and better alignment performance on unseen MMLU tasks than the calibration-only baseline.
• Score: 70.05453324928955
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Honesty alignment-the ability of large language models (LLMs) to recognize their knowledge boundaries and express calibrated confidence-is essential for trustworthy deployment. Existing methods either rely on training-free confidence estimation (e.g., token probabilities, self-consistency) or training-based calibration with correctness annotations. While effective, achieving universal honesty alignment with training-based calibration requires costly, large-scale labeling. To support annotation-efficient training, we introduce Elicitation-Then-Calibration (EliCal), a two-stage framework that first elicits internal confidence using inexpensive self-consistency supervision, then calibrates this confidence with a small set of correctness annotations. To support a large-scale study, we release HonestyBench, a benchmark covering ten free-form QA datasets with 560k training and 70k evaluation instances annotated with correctness and self-consistency signals. Experiments show that EliCal achieves near-optimal alignment with only 1k correctness annotations (0.18% of full supervision) and better alignment performance on unseen MMLU tasks than the calibration-only baseline, offering a scalable solution toward universal honesty alignment in LLMs.

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)
• CritiCal: Can Critique Help LLM Uncertainty or Confidence Calibration? [55.32645640455462]
  This paper studies how natural language critiques can enhance verbalized confidence.<n>We propose Self-Critique, enabling LLMs to critique and optimize their confidence beyond mere accuracy.<n> Experiments show that CritiCal significantly outperforms Self-Critique and other competitive baselines.
  arXiv  Detail & Related papers  (2025-10-28T15:16:06Z)
• ConfTuner: Training Large Language Models to Express Their Confidence Verbally [58.63318088243125]
  Large Language Models (LLMs) are increasingly deployed in high-stakes domains such as science, law, and healthcare.<n>LLMs are often observed to generate incorrect answers with high confidence, a phenomenon known as "overconfidence"
  arXiv  Detail & Related papers  (2025-08-26T09:25:32Z)
• Beyond Binary Rewards: Training LMs to Reason About Their Uncertainty [59.97939500426759]
  This paper describes RLCR, an approach to training reasoning models that jointly improves accuracy and confidence estimation.<n>We show that across diverse datasets, RLCR substantially improves calibration with no loss in accuracy.<n>We also demonstrate that verbalized confidence can be leveraged at test time to improve accuracy and calibration.
  arXiv  Detail & Related papers  (2025-07-22T17:56:01Z)
• Rewarding Doubt: A Reinforcement Learning Approach to Calibrated Confidence Expression of Large Language Models [34.59785123314865]
  A safe and trustworthy use of Large Language Models (LLMs) requires an accurate expression of confidence in their answers.<n>We propose a novel Reinforcement Learning approach that allows to directly fine-tune LLMs to express calibrated confidence estimates alongside their answers to factual questions.
  arXiv  Detail & Related papers  (2025-03-04T13:48:50Z)
• Fact-Level Confidence Calibration and Self-Correction [64.40105513819272]
  We propose a Fact-Level framework that calibrates confidence to relevance-weighted correctness at the fact level. We also develop Confidence-Guided Fact-level Self-Correction ($textbfConFix$), which uses high-confidence facts within a response as additional knowledge to improve low-confidence ones.
  arXiv  Detail & Related papers  (2024-11-20T14:15:18Z)
• Graph-based Confidence Calibration for Large Language Models [22.394717844099684]
  We propose using an auxiliary learning model to assess response correctness based on the self-consistency of multiple outputs generated by the large language models.<n>Our method builds a consistency graph to represent the agreement among multiple responses and uses a graph neural network (GNN) to estimate the likelihood that each response is correct.
  arXiv  Detail & Related papers  (2024-11-03T20:36:44Z)
• Consistency Calibration: Improving Uncertainty Calibration via Consistency among Perturbed Neighbors [22.39558434131574]
  We introduce the concept of consistency as an alternative perspective on model calibration. We propose a post-hoc calibration method called Consistency (CC) which adjusts confidence based on the model's consistency across inputs. We show that performing perturbations at the logit level significantly improves computational efficiency.
  arXiv  Detail & Related papers  (2024-10-16T06:55:02Z)
• Calibrating Large Language Models with Sample Consistency [76.23956851098598]
  We explore the potential of deriving confidence from the distribution of multiple randomly sampled model generations, via three measures of consistency. Results show that consistency-based calibration methods outperform existing post-hoc approaches. We offer practical guidance on choosing suitable consistency metrics for calibration, tailored to the characteristics of various LMs.
  arXiv  Detail & Related papers  (2024-02-21T16:15:20Z)
• Calibrating Long-form Generations from Large Language Models [34.72041258464477]
  Large Language Models' (LLMs) confidence scores should align with the actual likelihood of its responses being correct. Current confidence elicitation methods and calibration metrics rely on a binary true/false assessment of response correctness. We introduce a unified calibration framework, in which both the correctness of the LLMs' responses and their associated confidence levels are treated as distributions across a range of scores.
  arXiv  Detail & Related papers  (2024-02-09T17:00:32Z)
• Calibration by Distribution Matching: Trainable Kernel Calibration Metrics [56.629245030893685]
  We introduce kernel-based calibration metrics that unify and generalize popular forms of calibration for both classification and regression. These metrics admit differentiable sample estimates, making it easy to incorporate a calibration objective into empirical risk minimization. We provide intuitive mechanisms to tailor calibration metrics to a decision task, and enforce accurate loss estimation and no regret decisions.
  arXiv  Detail & Related papers  (2023-10-31T06:19:40Z)

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.