paper-with-me

홈 › Papers

Token-Level Uncertainty Estimation for Large Language Model Reasoning

2025-05-16 · Tunyu Zhang, Haizhou Shi, Yibin Wang, Hengyi Wang, Xiaoxiao He, Zhuowei Li, Haoxian Chen, Ligong Han, Kai Xu, huan zhang, Dimitris Metaxas, Hao Wang

While Large Language Models (LLMs) have demonstrated impressive capabilities, their output quality remains inconsistent across various application scenarios, making it difficult to identify trustworthy responses, especially in complex tasks requiring multi-step reasoning. In this paper, we propose a token-level uncertainty estimation framework to enable LLMs to self-assess and self-improve their generation quality in mathematical reasoning. Specifically, we introduce low-rank random weight perturbation to LLM decoding, generating predictive distributions that we use to estimate token-level uncertainties. We then aggregate these uncertainties to reflect semantic uncertainty of the generated sequences. Experiments on mathematical reasoning datasets of varying difficulty demonstrate that our token-level uncertainty metrics strongly correlate with answer correctness and model robustness. Additionally, we explore using uncertainty to directly enhance the model's reasoning performance through multiple generations and the particle filtering algorithm. Our approach consistently outperforms existing uncertainty estimation methods, establishing effective uncertainty estimation as a valuable tool for both evaluating and improving reasoning generation in LLMs.

📄 PDF Abstract BibTeX arXiv:2505.11737

Code (0)

등록된 구현이 없습니다.

Tasks

Language ModelingLanguage ModellingLarge Language ModelMathematical Reasoning

Similar Papers 제목 키워드 기반

Leveraging Visual Signals for Robust Token-Level Uncertainty in Vision-Language Generation

2026-05-26 · Joseph Hoche, David Brellmann, Gianni Franchi arxiv

Uncertainty quantification (UQ) remains a critical challenge in Large Vision Language Models (LVLMs) for reliable predictions and real-world deployment. However, most existing methods are adapted from the LLM literature …

Visual Grounding

GENUINE: Graph Enhanced Multi-level Uncertainty Estimation for Large Language Models

2025-09-09 · Tuo Wang, Adithya Kulkarni, Tyler Cody, Peter A. Beling 외 arxiv

Uncertainty estimation is essential for enhancing the reliability of Large Language Models (LLMs), particularly in high-stakes applications. Existing methods often overlook semantic dependencies, relying on token-level p…

Integrating Local and Global Entropy for Uncertainty Quantification in LLMs

2026-06-02 · Johanne Medina, Tianyi Zhou, Keivin Isufaj, Aristides Gionis 외 arxiv

Large language models hallucinate confidently, making uncertainty quantification (UQ) essential for reliable deployment. Existing methods rely predominantly on token-level signals, leaving the geometric structure of inte…

Uncertainty Estimation in Autoregressive Structured Prediction

2020-02-18 · ICLR 2021 1 · Andrey Malinin, Mark Gales

Uncertainty estimation is important for ensuring safety and robustness of AI systems. While most research in the area has focused on un-structured prediction tasks, limited work has investigated general uncertainty estim…

Image ClassificationMachine TranslationPredictionspeech-recognition+3

SpanUQ: Span-Level Uncertainty Quantification for Large Language Model Generation

2026-07-07 · Yimeng Zhang, Yingying Zhuang, Ziyi Wang, Yuxuan Lu 외 arxiv

Uncertainty estimation is essential not only for the trustworthy deployment of large language models (LLMs) but also as a foundation for self-refinement in LLM generation. However, existing approaches operate at suboptim…