5arXiv cs.CL (Computation and Language)·
Benchmarking LLM judges for citation verification in deep-research systems
A new arXiv paper evaluates 8 LLM judges from 3 model families on citation quality assessment for deep-research systems, testing across 1,248 rubric decisions with human-reviewed gold labels. The study finds that cheaper models remain competitive with frontier models — GPT-5-mini achieves the strongest source-relevance F1 at 0.908 — but judges differ substantially in directional bias (pass-rate drift, false positive/negative rates) even when scalar F1 scores are similar. The key finding is that scalar F1 obscures biases that would be directly reinforced in an RL training loop, making judge calibration a prerequisite before using citation rubrics as reward signals.
Related guides (3)
Related events (8)
Auditing LLM-as-Judge reliability: judge upgrades are not interchangeable across model families
A new arXiv paper investigates measurement validity problems in LLM-as-judge evaluation, finding that swapping evaluator models changes scores even when candidate responses are fixed. Across four judgment datasets, the authors compare Qwen3 dense judges (1.7B–32B) and MiniMax M2/M2.7 API releases, finding that only the Qwen3 1.7B→4B upgrade yields robust adjacent gains while MiniMax adjacent releases do not. Stronger judges reduce but do not eliminate position and verbosity bias, and repeated-sample juries add little when errors are correlated. The paper argues for standardized reporting requirements including dataset slices, bias probes, error-dependence estimates, and protocol audit trails.
LLMs judge worse than they generate: empirical challenge to self-evaluation pipeline assumptions
A new arXiv preprint tests the implicit assumption that LLM evaluation is easier than generation, using a controlled in-context QA setup across four benchmarks (SQuAD 2.0, DROP, HotpotQA, MuSiQue) and two models. Results show generation accuracy exceeds self-evaluation accuracy on three of four benchmarks, with attention analysis revealing that evaluation attends to context 3–5x less than generation does. LoRA fine-tuning experiments confirm the asymmetry is not a training artifact, with cross-task interference observed in both directions. The findings directly challenge assumptions underlying LLM-as-a-Judge and self-evaluation pipelines widely used in RLHF and agentic systems.
BabelJudge: Benchmark for measuring LLM-as-a-judge reliability across languages and agent trajectories
BabelJudge is a new open-source benchmark and audit framework that systematically measures four failure modes in LLM-as-a-judge systems: position bias, verbosity bias, order inconsistency, and cross-lingual degradation. The framework uses a 'gold-labelling by degradation' technique to generate labeled evaluation pairs without human annotation. Evaluation of Qwen2.5-7B-Instruct-4bit across English, Hindi, Arabic, and Swahili reveals severe cross-lingual reliability drops, with Swahili order consistency collapsing to near-random (0.480). The framework is extended to agentic evaluation with nine trajectory-level perturbations and three new metrics, released as a Python package supporting 11 judge backends.
Judge Arena: Benchmarking LLMs as Evaluators
Hugging Face and Atla have launched Judge Arena, a platform for benchmarking large language models in their role as automated evaluators. The initiative uses an Elo-based ranking system to compare how well different LLMs judge the quality of model outputs, addressing the growing reliance on LLM-as-judge paradigms in evaluation pipelines. This fills a meta-evaluation gap: as LLM judges become standard practice, understanding their relative reliability and biases becomes critical infrastructure for the field.
Survey and recommendations for LLM-as-a-Judge evaluation in multilingual and low-resource language settings
A new arXiv paper surveys 650 ACL Anthology papers that use LLM-as-a-Judge evaluation, finding only 33 address multilingual or low-resource language settings. Analysis of those 33 papers reveals inconsistent outcomes, overtrust in LLM judgments, and over-reliance on single judge models. The authors provide recommendations for improving evaluation practice in these underserved settings.
Paper challenges LLM expert-level claims by measuring variance and error magnitude in code-based data analysis tasks
A new arXiv paper argues that standard LLM benchmarks overstate model capabilities by focusing on average performance on training-data-adjacent tasks while ignoring response variance and error magnitude. The authors introduce a novel benchmark requiring frontier LLMs to write code for data analysis tasks, comparing results against human expert submissions. Human experts outperformed the frontier LLM on average across multiple metrics and showed lower performance variability. The findings challenge the prevailing narrative that LLMs perform at human-expert level on knowledge economy tasks.
Systematic study of LLM data referencing errors in tables, with lightweight critic model mitigation
A new arXiv paper introduces the first systematic evaluation of data referencing errors (DREs) — incorrect citation or omission of table values — across LLMs ranging from 1.7B to 20B parameters. The authors find DREs are pervasive across all tested models and tasks, compromising intermediate reasoning steps beyond just final-answer accuracy. They demonstrate that a critic-based filtering and rejection sampling approach improves answer accuracy by up to 12%, and train a lightweight 4B critic model achieving 78.2% F1 on detecting DREs both in- and out-of-distribution.
SoundnessBench: Benchmarking LLMs as Evaluators of ML Research Proposal Viability
SoundnessBench is a new benchmark of 1,099 machine-learning research proposals derived from ICLR submissions, labeled with reviewer soundness scores, designed to test whether LLMs can reliably distinguish methodologically sound research ideas from unsound ones. Evaluated across 12 frontier LLMs, the benchmark reveals a pervasive optimism bias: models systematically rate low-soundness proposals as sound under standard prompting, with aggressive prompting shifting errors from false positives to false negatives rather than eliminating them. Controls for data contamination, surface features, and human audit quality suggest the bias is not attributable to a single confounder. The authors conclude that current LLMs are not yet reliable as standalone first-gate evaluators of scientific rigor, a critical bottleneck for autonomous AI research agents.