Saved in:
Bibliographic Details
Main Authors: Nguyen, Manh, Gupta, Sunil, Le, Hung
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.26849
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916049452007424
author Nguyen, Manh
Gupta, Sunil
Le, Hung
author_facet Nguyen, Manh
Gupta, Sunil
Le, Hung
contents Sampling multiple responses improves language model reasoning, but uniform compute allocation is inefficient: easy questions are over-sampled while hard questions remain under-explored. We propose Uncertainty-Aware Budget Allocation (UAB), a concave integer optimization framework that reallocates a fixed sampling budget based on per-question uncertainty estimated at no additional inference cost. In Phase 1, every question receives one generation; its average negative log-likelihood (ANLL), extracted directly from output log-probabilities, serves as a difficulty signal while the generation contributes to the final vote. In Phase 2, the remaining budget is allocated by a marginal-greedy algorithm that solves a concave coverage-maximization surrogate exactly: uncertain questions receive more sampling budget while confident questions receive fewer additional samples. Evaluated on six open-weight and black-box models spanning 1.5B to 27B parameters and five reasoning benchmarks covering math, logic, and preference tasks, UAB outperforms baselines by up to +3% in average accuracy and up to +5% on individual benchmarks, with the largest gains in low-resource settings, requiring no auxiliary model or additional LLM call. Code is publicly available at https://github.com/manhitv/UAB.
format Preprint
id arxiv_https___arxiv_org_abs_2605_26849
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Uncertainty-Aware Budget Allocation for Adaptive Test-Time Reasoning
Nguyen, Manh
Gupta, Sunil
Le, Hung
Computation and Language
Sampling multiple responses improves language model reasoning, but uniform compute allocation is inefficient: easy questions are over-sampled while hard questions remain under-explored. We propose Uncertainty-Aware Budget Allocation (UAB), a concave integer optimization framework that reallocates a fixed sampling budget based on per-question uncertainty estimated at no additional inference cost. In Phase 1, every question receives one generation; its average negative log-likelihood (ANLL), extracted directly from output log-probabilities, serves as a difficulty signal while the generation contributes to the final vote. In Phase 2, the remaining budget is allocated by a marginal-greedy algorithm that solves a concave coverage-maximization surrogate exactly: uncertain questions receive more sampling budget while confident questions receive fewer additional samples. Evaluated on six open-weight and black-box models spanning 1.5B to 27B parameters and five reasoning benchmarks covering math, logic, and preference tasks, UAB outperforms baselines by up to +3% in average accuracy and up to +5% on individual benchmarks, with the largest gains in low-resource settings, requiring no auxiliary model or additional LLM call. Code is publicly available at https://github.com/manhitv/UAB.
title Uncertainty-Aware Budget Allocation for Adaptive Test-Time Reasoning
topic Computation and Language
url https://arxiv.org/abs/2605.26849