Research

Paper

AI LLM February 19, 2026

Fine-Grained Uncertainty Quantification for Long-Form Language Model Outputs: A Comparative Study

Authors

Dylan Bouchard, Mohit Singh Chauhan, Viren Bajaj, David Skarbrevik

Abstract

Uncertainty quantification has emerged as an effective approach to closed-book hallucination detection for LLMs, but existing methods are largely designed for short-form outputs and do not generalize well to long-form generation. We introduce a taxonomy for fine-grained uncertainty quantification in long-form LLM outputs that distinguishes methods by design choices at three stages: response decomposition, unit-level scoring, and response-level aggregation. We formalize several families of consistency-based black-box scorers, providing generalizations and extensions of existing methods. In our experiments across multiple LLMs and datasets, we find 1) claim-response entailment consistently performs better or on par with more complex claim-level scorers, 2) claim-level scoring generally yields better results than sentence-level scoring, and 3) uncertainty-aware decoding is highly effective for improving the factuality of long-form outputs. Our framework clarifies relationships between prior methods, enables apples-to-apples comparisons, and provides practical guidance for selecting components for fine-grained UQ.

Metadata

arXiv ID: 2602.17431
Provider: ARXIV
Primary Category: cs.CL
Published: 2026-02-19
Fetched: 2026-02-21 18:51

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2602.17431v1</id>\n    <title>Fine-Grained Uncertainty Quantification for Long-Form Language Model Outputs: A Comparative Study</title>\n    <updated>2026-02-19T15:02:29Z</updated>\n    <link href='https://arxiv.org/abs/2602.17431v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2602.17431v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Uncertainty quantification has emerged as an effective approach to closed-book hallucination detection for LLMs, but existing methods are largely designed for short-form outputs and do not generalize well to long-form generation. We introduce a taxonomy for fine-grained uncertainty quantification in long-form LLM outputs that distinguishes methods by design choices at three stages: response decomposition, unit-level scoring, and response-level aggregation. We formalize several families of consistency-based black-box scorers, providing generalizations and extensions of existing methods. In our experiments across multiple LLMs and datasets, we find 1) claim-response entailment consistently performs better or on par with more complex claim-level scorers, 2) claim-level scoring generally yields better results than sentence-level scoring, and 3) uncertainty-aware decoding is highly effective for improving the factuality of long-form outputs. Our framework clarifies relationships between prior methods, enables apples-to-apples comparisons, and provides practical guidance for selecting components for fine-grained UQ.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.CL'/>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.AI'/>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.LG'/>\n    <published>2026-02-19T15:02:29Z</published>\n    <arxiv:comment>UQLM repository: https://github.com/cvs-health/uqlm</arxiv:comment>\n    <arxiv:primary_category term='cs.CL'/>\n    <author>\n      <name>Dylan Bouchard</name>\n    </author>\n    <author>\n      <name>Mohit Singh Chauhan</name>\n    </author>\n    <author>\n      <name>Viren Bajaj</name>\n    </author>\n    <author>\n      <name>David Skarbrevik</name>\n    </author>\n  </entry>"
}