Confidence-Weighted Token Set Cover for Early Hypothesis Pruning in Self-Consistency

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Sultan, Md Arafat, Astudillo, Ramón Fernandez
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911093716156416
author Sultan, Md Arafat
Astudillo, Ramón Fernandez
author_facet Sultan, Md Arafat
Astudillo, Ramón Fernandez
contents Despite its simplicity and efficacy, the high token expenditure of self-consistency can limit its practical utility. Here we investigate if self-consistency can be made more token-efficient for long chain-of-thought reasoning tasks, while preserving its parallelism, through early hypothesis pruning. Concretely, we generate all solutions in parallel, but periodically prune intermediate hypotheses that are deemed unnecessary based on two lightweight indicators: (a) the model's own confidence in individual hypotheses, and (b) lexical coverage of all current hypotheses by candidate subsets that are under consideration for continued retention. We design a fast weighted set cover algorithm that utilizes the two indicators; our evaluation of five LLMs on three math benchmarks shows that this method can improve token efficiency for all models, by 10-35% in many cases.
format Preprint
id arxiv_https___arxiv_org_abs_2508_03979
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Confidence-Weighted Token Set Cover for Early Hypothesis Pruning in Self-Consistency
Sultan, Md Arafat
Astudillo, Ramón Fernandez
Computation and Language
Despite its simplicity and efficacy, the high token expenditure of self-consistency can limit its practical utility. Here we investigate if self-consistency can be made more token-efficient for long chain-of-thought reasoning tasks, while preserving its parallelism, through early hypothesis pruning. Concretely, we generate all solutions in parallel, but periodically prune intermediate hypotheses that are deemed unnecessary based on two lightweight indicators: (a) the model's own confidence in individual hypotheses, and (b) lexical coverage of all current hypotheses by candidate subsets that are under consideration for continued retention. We design a fast weighted set cover algorithm that utilizes the two indicators; our evaluation of five LLMs on three math benchmarks shows that this method can improve token efficiency for all models, by 10-35% in many cases.
title Confidence-Weighted Token Set Cover for Early Hypothesis Pruning in Self-Consistency
topic Computation and Language
url https://arxiv.org/abs/2508.03979