The paper examines LLM-as-a-Judge systems used to assess AI-generated text, questioning the assumption that judgments are derived from reasoning over responses and rubrics. It finds that classifiers trained solely on rubric text can predict judge outputs, indicating that rubrics contain recoverable evaluative signals independent of the responses. Counterfactual experiments show judges often fail to adjust decisions when either the response or rubric criterion is reversed, raising doubts about the reliability of rubric-based LLM evaluation.
By Anshul Bagaria, Sowmya S Sundaram, Gokul S Krishnan, Balaraman Ravindran
The paper introduces behavioral correctness assumptions as a new framework for evaluating reference-based automatic evaluation methods in natural language generation. It defines a taxonomy of correctness-preserving and correctness-altering assumptions and operationalizes them through controlled response transformations to specify expected scoring behaviors. The authors evaluate a range of lexical, character-level, semantic, LLM-based, and hybrid evaluators, analyzing their behavior across multiple dimensions and finding that no evaluator satisfies all assumptions, revealing distinct behavioral trade-offs not evident from aggregate scores.
By Maria Mahbub, Ashley Rice, Michael R. Munroe, Amidu Kamara, Amir Sadovnik
The paper introduces LongJudgeBench, a benchmark designed to evaluate large language models (LLMs) acting as judges for long-form text generation. It highlights that long-form evaluation requires complex, document-level assessments beyond simple length, such as organization, coverage, depth, consistency, and scenario-specific quality. Experiments show a significant reliability gap among current LLM judges, indicating instability across scenarios and limited effectiveness of rubrics or references.
By Junjie Chen, Yuxi Dong, Haitao Li, Weihang Su, Yujia Zhou, Min Zhang, Yiqun Liu, Qingyao Ai
arXiv:2606. 27226v1 Announce Type: new Abstract: Evaluating LLM outputs remains a major bottleneck in NLP: human evaluation is expensive and slow, lexical metrics correlate poorly with human judgments on open-ended generation, and holistic LLM judges often produce opaque scores that are hard to debug.
By Sangwoo Cho, Kushal Chawla, Pengshan Cai, Zefang Liu, Chenyang Zhu, Shi-Xiong Zhang, Sambit Sahu
The paper introduces UPHELD, a large benchmark of human-to-human dialogues written by professional script writers, featuring realistic turn densities and over 36,000 per-turn human annotations. It evaluates existing automatic metrics and LLM-as-a-judge methods, finding them unreliable against expert human judgment. Using UPHELD, the authors develop a Mixture-of-Judges framework that improves correlation with human assessments by about 30%.
By Ilija Subasic, Andrew Rabinovich, Zhao Chen
The paper introduces JudgeBiasBench, a benchmark that systematically quantifies judgment biases in large language model (LLM)-based judges across four dimensions and 12 bias types. It evaluates both generative and discriminative judges, revealing significant bias patterns that undermine reliability. The authors propose bias-aware training—reinforcement learning for generative judges and contrastive learning for discriminative judges—to reduce these biases while maintaining evaluation performance.
By Hongli Zhou, Hui Huang, Rui Zhang, Kehai Chen, Bing Xu, Conghui Zhu, Tiejun Zhao, Muyun Yang
Evaluating LLM outputs remains a major bottleneck in NLP: human evaluation is expensive and slow, lexical metrics correlate poorly with human judgments on open-ended generation, and holistic LLM judges often produce opaque scores that are hard to debug. We propose BINEVAL, a framework that decomposes evaluation criteria into atomic binary questions and aggregates the resulting verdicts into interpretable, multi-dimensional scores.
The paper "What Do We Expect from LLMs? Mapping the Design of LLM Benchmarks" analyzes 14,767 arXiv submissions from 2022 to 2026 that introduce or update evaluation resources for large language models. It systematically maps changes in target systems, domains, evaluation materials, conditions, and scoring mechanisms, revealing a growing emphasis on action, interaction, and professional applications. The study also notes uneven development in model participation, with LLM-based scoring increasing in both agent and non-agent groups, while model-generated materials do not show a comparable rise.
By Chao Wang (Independent Researcher)
arXiv:2609.14819v1 Announce Type: cross
Abstract: Large language models (LLMs) have a growing range of applications in medicine, and their evaluation is critical for ensuring they provide benefit and...
By Suzannah E McKinney, Phuc Vu, Samuel A Justice, Christopher Humphries, Alyssa Pradhan, Timothy J Keyes, Bernardo C Bizzo, Keith J Dreyer, Sarah F Mercaldo, James M Hillis
arXiv:2609.15561v1 Announce Type: new
Abstract: Evaluating the factual correctness of large language models (LLMs) is vital for many applications. But are our evaluation tools themselves trustworthy?...
By Sarra Gharsallah, Adele Robaldo, Mariia Tokareva, Giovanni Gatti Pinheiro, Ilyana Guendouz, Rapha\"el Troncy, Paolo Papotti, Pietro Michiardi
arXiv:2601.08654v3 Announce Type: replace
Abstract: Rubric-based text evaluation increasingly relies on large language models (LLMs) as scalable judges, yet frozen black-box models can interpret the...
By Yihan Hong, Huaiyuan Yao, Bolin Shen, Wanpeng Xu, Hua Wei, Yushun Dong
CriticGen introduces a generation‑aware evaluation framework that generates sample‑specific evaluation dimensions and scoring criteria across categories such as subjective, objective, and self‑derived constraints. These dynamic rubrics produce a score, reason, executable refinement suggestion, and a refined answer, enabling models to diagnose and target flaws in their responses. Experiments show significant gains in rubric quality, score correlation, and actionable feedback, with 73.17% of answers improved and a 93.28% non‑degradation rate.
By Huifang Du, Zecheng Zuo, Sen Wang, Chenghao Fan, Haofen Wang, Yehui Yang