A Theoretical Study on Bridging Internal Probability and Self-Consistency for LLM Reasoning

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Zhou, Zhi, Tan, Yuhao, Li, Zenan, Yao, Yuan, Guo, Lan-Zhe, Li, Yu-Feng, Ma, Xiaoxing
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866908599432773632
author Zhou, Zhi
Tan, Yuhao
Li, Zenan
Yao, Yuan
Guo, Lan-Zhe
Li, Yu-Feng
Ma, Xiaoxing
author_facet Zhou, Zhi
Tan, Yuhao
Li, Zenan
Yao, Yuan
Guo, Lan-Zhe
Li, Yu-Feng
Ma, Xiaoxing
contents Test-time scaling seeks to improve the reasoning performance of large language models (LLMs) by adding computational resources. A prevalent approach within the field is sampling-based test-time scaling methods, which enhance reasoning by generating multiple reasoning paths for a given input during inference. However, despite its practical success, the theoretical foundations remain underexplored. In this paper, we provide the first theoretical framework for analyzing sampling-based test-time scaling methods, grounded in the perspective of confidence estimation. Based on the framework, we analyze two dominant paradigms: self-consistency and perplexity, and reveal key limitations: self-consistency suffers from high estimation error while perplexity exhibits substantial modeling error and possible degradation of the estimation error convergence. To address these limitations, we introduce RPC, a hybrid method that leverages our theoretical insights through two key components: Perplexity Consistency and Reasoning Pruning. Perplexity Consistency combines the strengths of self-consistency and perplexity, boosting the convergence rate of estimation error from linear to exponential while preserving model error. Reasoning Pruning prevents degradation by eliminating low-probability reasoning paths. Both theoretical analysis and empirical results across seven benchmark datasets demonstrate that RPC has a strong potential for reducing reasoning error. Notably, RPC achieves reasoning performance comparable to self-consistency while not only enhancing confidence reliability but also reducing sampling costs by 50%. The code and resources are available at https://wnjxyk.github.io/RPC.
format Preprint
id arxiv_https___arxiv_org_abs_2510_15444
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle A Theoretical Study on Bridging Internal Probability and Self-Consistency for LLM Reasoning
Zhou, Zhi
Tan, Yuhao
Li, Zenan
Yao, Yuan
Guo, Lan-Zhe
Li, Yu-Feng
Ma, Xiaoxing
Machine Learning
Artificial Intelligence
Test-time scaling seeks to improve the reasoning performance of large language models (LLMs) by adding computational resources. A prevalent approach within the field is sampling-based test-time scaling methods, which enhance reasoning by generating multiple reasoning paths for a given input during inference. However, despite its practical success, the theoretical foundations remain underexplored. In this paper, we provide the first theoretical framework for analyzing sampling-based test-time scaling methods, grounded in the perspective of confidence estimation. Based on the framework, we analyze two dominant paradigms: self-consistency and perplexity, and reveal key limitations: self-consistency suffers from high estimation error while perplexity exhibits substantial modeling error and possible degradation of the estimation error convergence. To address these limitations, we introduce RPC, a hybrid method that leverages our theoretical insights through two key components: Perplexity Consistency and Reasoning Pruning. Perplexity Consistency combines the strengths of self-consistency and perplexity, boosting the convergence rate of estimation error from linear to exponential while preserving model error. Reasoning Pruning prevents degradation by eliminating low-probability reasoning paths. Both theoretical analysis and empirical results across seven benchmark datasets demonstrate that RPC has a strong potential for reducing reasoning error. Notably, RPC achieves reasoning performance comparable to self-consistency while not only enhancing confidence reliability but also reducing sampling costs by 50%. The code and resources are available at https://wnjxyk.github.io/RPC.
title A Theoretical Study on Bridging Internal Probability and Self-Consistency for LLM Reasoning
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2510.15444