Saved in:
Bibliographic Details
Main Authors: Chen, Yongrui, Ma, Yangyang, Huang, Xiaoying, Zhang, Shenyu, Chen, Huajun, Wang, Haofen, Qi, Guilin
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.01939
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918479657959424
author Chen, Yongrui
Ma, Yangyang
Huang, Xiaoying
Zhang, Shenyu
Chen, Huajun
Wang, Haofen
Qi, Guilin
author_facet Chen, Yongrui
Ma, Yangyang
Huang, Xiaoying
Zhang, Shenyu
Chen, Huajun
Wang, Haofen
Qi, Guilin
contents Static benchmarks for LLMs are increasingly compromised by contamination and overfitting especially on knowledge intensive reasoning tasks While recent dynamic benchmarks can alleviate staleness they often increase difficulty at the expense of answerability and controllability In this paper we propose StressEval a failure driven data synthesis framework that turns observed model failures into dynamic challenging and controllable test instances StressEval consists of three stages first it constructs a semi structured difficulty card that identifies the failed reasoning step and its root cause second it applies a dual perspective instance synthesis method that targets both knowledge gaps and reasoning breakdowns while preserving the underlying difficulty factors and third it applies a gating mechanism to retain only grounded unambiguous instances Seeding from multiple knowledge intensive reasoning datasets we employ StressEval to build Dynamic OneEval a focused suite of challenging dynamic benchmark Across several state of the art LLMs Dynamic OneEval yields substantially larger performance drops than the original benchmarks while retaining explicit difficulty factors enabling more actionable iteration
format Preprint
id arxiv_https___arxiv_org_abs_2605_01939
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle StressEval: Failure-Driven Dynamic Benchmarking for Knowledge-Intensive Reasoning in Large Language Models
Chen, Yongrui
Ma, Yangyang
Huang, Xiaoying
Zhang, Shenyu
Chen, Huajun
Wang, Haofen
Qi, Guilin
Computation and Language
Static benchmarks for LLMs are increasingly compromised by contamination and overfitting especially on knowledge intensive reasoning tasks While recent dynamic benchmarks can alleviate staleness they often increase difficulty at the expense of answerability and controllability In this paper we propose StressEval a failure driven data synthesis framework that turns observed model failures into dynamic challenging and controllable test instances StressEval consists of three stages first it constructs a semi structured difficulty card that identifies the failed reasoning step and its root cause second it applies a dual perspective instance synthesis method that targets both knowledge gaps and reasoning breakdowns while preserving the underlying difficulty factors and third it applies a gating mechanism to retain only grounded unambiguous instances Seeding from multiple knowledge intensive reasoning datasets we employ StressEval to build Dynamic OneEval a focused suite of challenging dynamic benchmark Across several state of the art LLMs Dynamic OneEval yields substantially larger performance drops than the original benchmarks while retaining explicit difficulty factors enabling more actionable iteration
title StressEval: Failure-Driven Dynamic Benchmarking for Knowledge-Intensive Reasoning in Large Language Models
topic Computation and Language
url https://arxiv.org/abs/2605.01939