Does Context Matter? ContextualJudgeBench for Evaluating LLM-based Judges in Contextual Settings

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Xu, Austin, Bansal, Srijan, Ming, Yifei, Yavuz, Semih, Joty, Shafiq
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916657866211328
author Xu, Austin
Bansal, Srijan
Ming, Yifei
Yavuz, Semih
Joty, Shafiq
author_facet Xu, Austin
Bansal, Srijan
Ming, Yifei
Yavuz, Semih
Joty, Shafiq
contents The large language model (LLM)-as-judge paradigm has been used to meet the demand for a cheap, reliable, and fast evaluation of model outputs during AI system development and post-deployment monitoring. While judge models -- LLMs finetuned to specialize in assessing and critiquing model outputs -- have been touted as general purpose evaluators, they are typically evaluated only on non-contextual scenarios, such as instruction following. The omission of contextual settings -- those where external information is used as context to generate an output -- is surprising given the increasing prevalence of retrieval-augmented generation (RAG) and summarization use cases. Contextual assessment is uniquely challenging, as evaluation often depends on practitioner priorities, leading to conditional evaluation criteria (e.g., comparing responses based on factuality and then considering completeness if they are equally factual). To address the gap, we propose ContextualJudgeBench, a judge benchmark with 2,000 challenging response pairs across eight splits inspired by real-world contextual evaluation scenarios. We build our benchmark with a multi-pronged data construction pipeline that leverages both existing human annotations and model-based perturbations. Our comprehensive study across 11 judge models and 9 general purpose models, reveals that the contextual information and its assessment criteria present a significant challenge to even state-of-the-art models. For example, OpenAI's o1, the best-performing model, barely reaches 55% consistent accuracy.
format Preprint
id arxiv_https___arxiv_org_abs_2503_15620
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Does Context Matter? ContextualJudgeBench for Evaluating LLM-based Judges in Contextual Settings
Xu, Austin
Bansal, Srijan
Ming, Yifei
Yavuz, Semih
Joty, Shafiq
Computation and Language
Artificial Intelligence
Machine Learning
The large language model (LLM)-as-judge paradigm has been used to meet the demand for a cheap, reliable, and fast evaluation of model outputs during AI system development and post-deployment monitoring. While judge models -- LLMs finetuned to specialize in assessing and critiquing model outputs -- have been touted as general purpose evaluators, they are typically evaluated only on non-contextual scenarios, such as instruction following. The omission of contextual settings -- those where external information is used as context to generate an output -- is surprising given the increasing prevalence of retrieval-augmented generation (RAG) and summarization use cases. Contextual assessment is uniquely challenging, as evaluation often depends on practitioner priorities, leading to conditional evaluation criteria (e.g., comparing responses based on factuality and then considering completeness if they are equally factual). To address the gap, we propose ContextualJudgeBench, a judge benchmark with 2,000 challenging response pairs across eight splits inspired by real-world contextual evaluation scenarios. We build our benchmark with a multi-pronged data construction pipeline that leverages both existing human annotations and model-based perturbations. Our comprehensive study across 11 judge models and 9 general purpose models, reveals that the contextual information and its assessment criteria present a significant challenge to even state-of-the-art models. For example, OpenAI's o1, the best-performing model, barely reaches 55% consistent accuracy.
title Does Context Matter? ContextualJudgeBench for Evaluating LLM-based Judges in Contextual Settings
topic Computation and Language
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2503.15620