Guardado en:
Detalles Bibliográficos
Autores principales: Saha, Aadirupa, Wagde, Aniket, Kveton, Branislav
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:https://arxiv.org/abs/2602.15481
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866914465657651200
author Saha, Aadirupa
Wagde, Aniket
Kveton, Branislav
author_facet Saha, Aadirupa
Wagde, Aniket
Kveton, Branislav
contents LLM-as-a-judge has emerged as a cornerstone technique for evaluating large language models by leveraging LLM reasoning to score prompt-response pairs. Since LLM judgments are stochastic, practitioners commonly query each pair multiple times to estimate mean scores accurately. This raises a critical challenge: given a fixed computational budget $B$, how to optimally allocate queries across $K$ prompt-response pairs to minimize estimation error? We present a principled variance-adaptive approach leveraging multi-armed bandit theory and concentration inequalities. Our method dynamically allocates queries based on estimated score variances, concentrating resources where uncertainty is highest. Further, our algorithm is shown to achieve a worst-case score-estimation error of $\tilde{O}\left(\sqrt{\frac{\sum_{i=1}^K σ_i^2}{B}}\right)$, $σ_i^2$ being the unknown score variance for pair $i \in [K]$ with near-optimal budget allocation. Experiments on \emph{Summarize-From-Feedback} and \emph{HelpSteer2} demonstrate that our method significantly outperforms uniform allocation, reducing worst-case estimation error while maintaining identical budgets. Our work establishes a theoretical foundation for efficient LLM evaluation with practical implications for AI safety, model alignment, and automated assessment at scale.
format Preprint
id arxiv_https___arxiv_org_abs_2602_15481
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle LLM-as-Judge on a Budget
Saha, Aadirupa
Wagde, Aniket
Kveton, Branislav
Machine Learning
LLM-as-a-judge has emerged as a cornerstone technique for evaluating large language models by leveraging LLM reasoning to score prompt-response pairs. Since LLM judgments are stochastic, practitioners commonly query each pair multiple times to estimate mean scores accurately. This raises a critical challenge: given a fixed computational budget $B$, how to optimally allocate queries across $K$ prompt-response pairs to minimize estimation error? We present a principled variance-adaptive approach leveraging multi-armed bandit theory and concentration inequalities. Our method dynamically allocates queries based on estimated score variances, concentrating resources where uncertainty is highest. Further, our algorithm is shown to achieve a worst-case score-estimation error of $\tilde{O}\left(\sqrt{\frac{\sum_{i=1}^K σ_i^2}{B}}\right)$, $σ_i^2$ being the unknown score variance for pair $i \in [K]$ with near-optimal budget allocation. Experiments on \emph{Summarize-From-Feedback} and \emph{HelpSteer2} demonstrate that our method significantly outperforms uniform allocation, reducing worst-case estimation error while maintaining identical budgets. Our work establishes a theoretical foundation for efficient LLM evaluation with practical implications for AI safety, model alignment, and automated assessment at scale.
title LLM-as-Judge on a Budget
topic Machine Learning
url https://arxiv.org/abs/2602.15481