PADM'E is a method for synthesizing preference‑aligned data to meta‑evaluate language‑model (LM) evaluators of agentic behaviors. It reframes meta‑evaluation as a preference judgment problem, generating criterion‑based data with small LMs and no human involvement. In a prototype, PADM'E produced 1,000 samples across four domains and three criteria, and human validation showed agreement with human judgment rising from 73% to 85% compared to a naive baseline.
By Cheng Chang, Yining Mao, Peng Qi
Large Language Model judges are commonly used to rank texts via pairwise comparison, with reliability traditionally measured by position bias, transitivity, and pairwise agreement. This paper argues that these proxies are misleading because they are dominated by close‑rank‑gap pairs, which contribute little to the overall ranking, while far‑gap pairs carry the true ranking signal. Experiments on simulations and human‑rated corpora show weak correlation between the proxies and actual ranking accuracy, suggesting judges should be evaluated using rank‑gap‑conditional metrics against human rankings.
By Bruno Brocai, Maria Becker
arXiv:2609.23201v1 Announce Type: new
Abstract: Benchmark scores increasingly influence the development, marketing, and selection of large language models (LLMs). Yet an overall score is interpretabl...
By Danial Amin
The paper argues that traditional global calibration metrics, such as Expected Calibration Error and Brier Score, are confounded by differences in model accuracy when comparing large language models. It introduces ACE, an accuracy‑controlled evaluation framework that offers Instance‑Aligned, Distribution‑Aligned, and Candidate‑Aligned views to provide fairer cross‑model comparisons. Experiments across various benchmarks reveal that many reported calibration advantages disappear after accuracy control and that model rankings often reverse, indicating that raw global metrics are unreliable for cross‑model calibration assessment.
By Zhichao Yang, Caiqi Zhang, Ruihan Yang, Chengzu Li, Nigel Collier, Deqing Yang
The paper demonstrates that large language model (LLM) evaluators, whether reward‑model based or prompted LLM‑as‑a‑Judge, exhibit significant language bias in multilingual settings. Experiments with semantically identical instruction‑response pairs across 23 languages reveal that lower‑resource languages receive higher scores, a bias that persists across eight open‑weight evaluators and is not detectable by standard pairwise accuracy metrics. The authors link the bias to model uncertainty and language identity, showing it cannot be explained by content difficulty alone.
By Ej Zhou, Lucas Resck, Zheng Hui, Anna Korhonen
arXiv:2607. 04581v1 Announce Type: cross Abstract: Text embeddings for Portuguese have no dedicated benchmark: evaluation rests on translated corpora such as English MS MARCO or on thin multilingual coverage, with native tasks scattered and unconsolidated.
By Tardelli Ronan Coelho Stekel
arXiv:2607. 04581v2 Announce Type: replace-cross Abstract: Text embeddings for Portuguese have no dedicated benchmark: evaluation rests on translated corpora such as English MS MARCO or on thin multilingual coverage, with native tasks scattered and unconsolidated.
By Tardelli Ronan Coelho Stekel
However, whether these judges truly evaluate the scientific substance of ideas or are influenced by superficial stylistic presentation remains an open question. To address this question, we propose SciStyleBench, a unified three-component benchmark for diagnosing and mitigating stylistic bias in LLM-based idea evaluation: (i) First, SciStyleStage, a three-stage evaluation environment that applies controlled stylistic perturbations to fixed scientific content across three settings no context, fixed-domain context, and open-domain retrieval context, covering 600 scientific ideas and 15 style variants, with 9,000 evaluation instances per setting; (ii) Second, SciStyleMetrics, a set of quantitative measures, including Style Bias Index (SBI), Substance Recognition Rate (SRR), and Adversarial Win Rate (AWR), to characterize how stylistic variation affects scoring stability, substance discrimination, and ranking robustness; (iii) Third, SciStyleExtractor, a plug-and-play evaluation module that separates presentation style from scientific content by predicting style type and deviation before style-conditioned evaluation, enabling us to assess whether style awareness reduces stylistic bias.
arXiv:2606. 13221v2 Announce Type: replace Abstract: Evaluating new large language models typically requires costly human annotation campaigns at scale.
By Bora Kargi, David Salinas
arXiv:2606. 09409v1 Announce Type: new Abstract: Pairwise comparisons combined with aggregation methods like Elo have become central to evaluating generative models, yet concerns remain that they reward superficial stylistic cues or display judge biases.
By Mina Remeli, Moritz Hardt
arXiv:2601. 21817v2 Announce Type: replace-cross Abstract: Evaluating large language models (LLMs) on open-ended tasks without ground-truth labels is increasingly done via the LLM-as-a-judge paradigm.
By Mingyuan Xu, Xinzi Tan, Jiawei Wu, Doudou Zhou
arXiv:2604. 07102v2 Announce Type: replace-cross Abstract: Activation-based steering enables inference-time personalization of large language models, but its effects in educational applications are not well understood.
By Yongchao Wu, Aron Henriksson