mlsanityData to 14 Oct 2026

Revisiting Uncertainty Quantification Evaluation in Language Models: Spurious Interactions with Response Length Bias Results

arXiv 2504.13677 ↗First posted 18 Apr 2025cs.CL cs.AI cs.LGcited by 21

Not on the map. Why?

Abstract (arXiv)

Uncertainty Quantification (UQ) in Language Models (LMs) is key to improving their safety and reliability. Evaluations often use metrics like AUROC to assess how well UQ methods (e.g., negative sequence probabilities) correlate with task correctness functions (e.g., ROUGE-L). We show that mutual biases--when both UQ methods and correctness functions are biased by the same factors--systematically distort evaluation. First, we formally prove that any mutual bias non-randomly skews AUROC rankings, compromising benchmark integrity. Second, we confirm this happens empirically by testing 7 widely used correctness functions, from lexical-based and embedding-based metrics to LM-as-a-judge approaches, across 4 datasets x 4 models x 8 UQ methods. Our analysis shows that length biases in correctness functions distort UQ assessments by interacting with length biases in UQ methods. We identify LM-as-a-judge methods as the least length-biased, offering a promising path for a fairer UQ evaluation.

Key facts. Cited by 21 papers in the map (data through 14 Oct 2026). Builds on 27 of its 41 references that are in the map. First posted on arXiv 18 Apr 2025.

Builds on

27 of 41 references.

References · 27

  1. 2019RoBERTa: A Robustly Optimized BERT Pretraining Approachcited by 13,372
  2. 2019Sentence-BERT: Sentence Embeddings using Siamese BERT-Networkscited by 9,170
  3. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,626
  4. 2019BERTScore: Evaluating Text Generation with BERTcited by 5,512
  5. 2024Qwen2.5 Technical Reportcited by 4,942
  6. 2016SQuAD: 100,000+ Questions for Machine Comprehension of Textcited by 4,148
  7. 2017TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehensioncited by 2,955
  8. 2022Language Models (Mostly) Know What They Knowcited by 1,856
  9. 2023Large Language Models are not Fair Evaluatorscited by 1,085
  10. 2019Latent Retrieval for Weakly Supervised Open Domain Question Answeringcited by 695
  11. 2020SummEval: Re-evaluating Summarization Evaluationcited by 674
  12. 2024Measuring short-form factuality in large language modelscited by 373
15 more
  1. 2024INSIDE: LLMs' Internal States Retain the Power of Hallucination Detectioncited by 323
  2. 2024Humans or LLMs as the Judge? A Study on Judgement Biasescited by 323
  3. 2023AlignScore: Evaluating Factual Consistency with a Unified Alignment Functioncited by 288
  4. 2024LLMs instead of Human Judges? A Large Scale Empirical Study across 20 NLP Evaluation Taskscited by 243
  5. 2020Unsupervised Quality Estimation for Neural Machine Translationcited by 211
  6. 2021On Hallucination and Predictive Uncertainty in Conditional Language Generationcited by 176
  7. 2023LM-Polygraph: Uncertainty Estimation for Language Modelscited by 131
  8. 2023Hallucinations in Large Multilingual Translation Modelscited by 129
  9. 2023Uncertainty in Natural Language Generation: From Theory to Applicationscited by 61
  10. 2024Linguistic Calibration of Long-Form Generationscited by 60
  11. 2024Improving Uncertainty Estimation through Semantically Diverse Language Generationcited by 35
  12. 2024Style Outweighs Substance: Failure Modes of LLM Judges in Alignment Benchmarkingcited by 35
  13. 2016Revisiting Summarization Evaluation for Scientific Articlescited by 31
  14. 2022Evaluate & Evaluation on the Hub: Better Best Practices for Data and Model Measurementscited by 22
  15. 2024LoGU: Long-form Generation with Uncertainty Expressionscited by 19

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 8; 2026: 13.

21 papers.

Sort:

By area:

  1. 2025Geometric Uncertainty for Detecting and Correcting Hallucinations in LLMscited by 13
  2. 2025When Judgment Becomes Noise: How Design Failures in LLM Judge Benchmarks Silently Undermine Validitycited by 11
  3. 2026Complementing Self-Consistency with Cross-Model Disagreement for Uncertainty Quantificationcited by 10
  4. 2025The Geometries of Truth Are Orthogonal Across Taskscited by 9
  5. 2025Addressing Pitfalls in the Evaluation of Uncertainty Estimation Methods for Natural Language Generationcited by 7
  6. 2025SelfReflect: Can LLMs Communicate Their Internal Answer Distribution?cited by 6
  7. 2025UNCERTAINTY-LINE: Length-Invariant Estimation of Uncertainty for Large Language Modelscited by 5
  8. 2026Semantic Self-Distillation for Language Model Uncertaintycited by 4
  9. 2026The Confidence Manifold: Geometric Structure of Correctness Representations in Language Modelscited by 4
  10. 2025Is Your Model Fairly Certain? Uncertainty-Aware Fairness Evaluation for LLMscited by 2
  11. 2026Improving Semantic Uncertainty Quantification in Language Model Question-Answering via Token-Level Temperature Scalingcited by 2
  12. 2026Spilled Energy in Large Language Modelscited by 1
9 more
  1. 2026A Robust Evaluation of Probe Robustness: Lessons for Reliable OOD Uncertainty Quantificationcited by 1
  2. 2025Transparent Reference-free Automated Evaluation of Open-Ended User Survey Responsescited by 0
  3. 2026EmplifAI: a Fine-grained Dataset for Japanese Empathetic Medical Dialogues in 28 Emotion Labelscited by 0
  4. 2026Useful Features, Backward Scores: OOD in Language-Model Trajectoriescited by 0
  5. 2026Self-Attention as Transport: Limits of Symmetric Spectral Diagnosticscited by 0
  6. 2026Position: Uncertainty Quantification in LLMs is Just Unsupervised Clusteringcited by 0
  7. 2026How Do Uncertainty Estimators Align with Hallucination Occurrence in LLMs?cited by 0
  8. 2026Estimating Uncertainty from Reasoning: A Large-Scale Study of Multi- and Crosslingual MCQA Performance in LLMscited by 0
  9. 2026Rethinking Correctness for Uncertainty Estimation in Clinical Prediction with Vision-Language Modelscited by 0