arXiv:2512. 03019v2 Announce Type: replace-cross Abstract: Thinking Large Language Models (LLMs) used as judges for pairwise preferences remain noisy at the single-sample level, and common aggregation rules (majority vote, soft self-consistency, or instruction-based self-aggregation) are inconsistent when ties are allowed.
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.
The paper argues that language models’ intransitive preferences arise from multiple internally consistent latent orderings rather than noise around a single ordering. By demonstrating that a single ordering cannot explain observed inconsistencies and introducing a noise‑augmented mixture Bradley‑Terry model, the authors show that mixtures of orderings better capture preference structure across several models and tasks. A case study on Moral Machine dilemmas further illustrates that models can share latent components even when aggregate preferences differ.
arXiv:2608. 02455v1 Announce Type: new Abstract: Human-centered assessment tasks, which are essential for systematic decision-making, rely heavily on human judgment and typically lack verifiable ground truth.
Human-centered assessment tasks, which are essential for systematic decision-making, rely heavily on human judgment and typically lack verifiable ground truth. Existing approaches face a dilemma: methods using only human judgments suffer from heterogeneous expertise and inconsistent rating scales, while methods using only model-generated scores must learn from imperfect proxies or incomplete features.
arXiv:2606. 13221v2 Announce Type: replace Abstract: Evaluating new large language models typically requires costly human annotation campaigns at scale.
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.
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.
SCOPE is a framework that calibrates an acceptance threshold for large language models used as pairwise judges, ensuring that the error rate among non-abstained judgments does not exceed a user-specified level α. It introduces Bidirectional Preference Entropy (BPE) to provide a bias-neutral uncertainty signal by querying the judge in both response positions and converting the averaged preference probability into an entropy-based score. Across multiple pairwise judging benchmarks, BPE outperforms standard confidence proxies in calibration and discrimination, while SCOPE consistently meets the target risk bound (empirical FDR ≈0.097–0.099 at α=0.10) and retains substantial coverage, accepting up to 2.4× more judgments under the same risk constraint.
arXiv:2601. 21816v2 Announce Type: replace Abstract: Evaluating the performance of large language models (LLMs) from human preference data is crucial for obtaining LLM leaderboards.
arXiv:2602.02219v3 Announce Type: replace Abstract: Large language models are widely employed as evaluators, a paradigm commonly referred to as LLM-as-a-judge. Prior research has predominantly examin...
arXiv:2604. 06996v2 Announce Type: replace-cross Abstract: LLM-as-a-judge has become the de facto approach for evaluating LLM outputs.
The paper critiques the conventional method of averaging human ratings to evaluate natural language generation (NLG) systems, arguing that it relies on assumptions about annotators that are often violated, especially when using Likert scales. These violations can even reverse true preferences, leading to inaccurate system rankings. The authors propose a more theoretically sound protocol and introduce a new system-level probabilistic assessment (SPA) for open-ended tasks like story generation, which successfully recovers expected model orderings where the standard protocol fails.