FollowTable is the first large-scale benchmark for instruction-following table retrieval, paired with an Instruction Responsiveness Score, showing that existing models fail to adapt to fine-grained constraints beyond topical similarity.
super hub Mixed citations
A Survey on LLM-as-a-Judge
Mixed citation behavior. Most common role is background (70%).
167
Pith papers citing it
Background
70% of classified citations
abstract
Accurate and consistent evaluation is crucial for decision-making across numerous fields, yet it remains a challenging task due to inherent subjectivity, variability, and scale. Large Language Models (LLMs) have achieved remarkable success across diverse domains, leading to the emergence of "LLM-as-a-Judge," where LLMs are employed as evaluators for complex tasks. With their ability to process diverse data types and provide scalable, cost-effective, and consistent assessments, LLMs present a compelling alternative to traditional expert-driven evaluations. However, ensuring the reliability of LLM-as-a-Judge systems remains a significant challenge that requires careful design and standardization. This paper provides a comprehensive survey of LLM-as-a-Judge, addressing the core question: How can reliable LLM-as-a-Judge systems be built? We explore strategies to enhance reliability, including improving consistency, mitigating biases, and adapting to diverse assessment scenarios. Additionally, we propose methodologies for evaluating the reliability of LLM-as-a-Judge systems, supported by a novel benchmark designed for this purpose. To advance the development and real-world deployment of LLM-as-a-Judge systems, we also discussed practical applications, challenges, and future directions. This survey serves as a foundational reference for researchers and practitioners in this rapidly evolving field.
hub tools
citation-role summary
background 18
method 5
citation-polarity summary
claims ledger
- abstract Accurate and consistent evaluation is crucial for decision-making across numerous fields, yet it remains a challenging task due to inherent subjectivity, variability, and scale. Large Language Models (LLMs) have achieved remarkable success across diverse domains, leading to the emergence of "LLM-as-a-Judge," where LLMs are employed as evaluators for complex tasks. With their ability to process diverse data types and provide scalable, cost-effective, and consistent assessments, LLMs present a compelling alternative to traditional expert-driven evaluations. However, ensuring the reliability of L
authors
co-cited works
representative citing papers
MathNet delivers the largest multilingual Olympiad math dataset and benchmarks where models like Gemini-3.1-Pro reach 78% on solving but embedding models struggle on equivalent problem retrieval, with retrieval augmentation yielding up to 12% gains.
GIANTS-4B, trained with RL on a new 17k-example benchmark of parent-to-child paper insights, achieves 34% relative improvement over gemini-3-pro in LM-judge similarity and is rated higher-impact by a citation predictor.
MediQAl is a new French medical QA benchmark with 32k exam-sourced questions in three formats and cognitive labels, evaluated on 14 LLMs to reveal gaps between factual recall and reasoning performance.
FARS deployed at scale produced 166 AI/ML papers across 67 topics that received 282 structured human reviews indicating some review-worthy outputs alongside recurring failure modes.
COHORT automates mitigation generation for network attacks via collaborative LLMs on emulated topologies with offensive replay evaluation, reporting 46.7% success rate that is 4.4 times higher than a single-agent baseline.
LLMs are applied in a generative pipeline for extracting, normalizing, and interpreting eligibility criteria from securities prospectuses, achieving up to 91% precision in document-level decisions with a conservative bias.
RealMath-Eval benchmark shows LLM judges have an evaluation gap, performing worse on diverse real human math reasoning than on synthetic solutions due to greater error diversity and higher surprisal.
ReasoningFlow represents LLM reasoning traces as DAGs, finding structural similarity across models and that most erroneous steps are unused in final answers.
CoEval generates task-specific benchmarks by rotating models through teacher, student, and judge roles, then weights questions by discriminative power and judges by panel consensus to recover accurate model rankings without labels.
Using frontier models to synthesize plausible-but-wrong FIM completions as hard negatives for SFT improves Delulu exact match by +18.8 and edit similarity by +0.22 on Qwen2.5-Coder-7B while also lifting HumanEval-Infilling and SAFIM.
Introduces the Matter to Mechanism benchmark of 2,645 structured instances and a composite metric suite for evaluating AI co-scientists on problem-to-hypothesis reasoning in battery materials research.
RWGBench is a citation-centric benchmark for related work generation built from 40k CS papers and a 100-paper test set, with multi-dimensional metrics that better match human expert judgment than standard similarity scores.
VIABLE benchmark reveals existing VLM judges are unreliable for VIA tasks (GPT-5.4 at 52.6% diagnostic accuracy with 94.2% self-preference) and proposes VIA-Judge-Agent for improvements.
DirectorBench is a profile-aware diagnostic benchmark that localizes bottlenecks in long-form video generation workflows using structured checkpoints and multi-agent evaluation.
Constructs KVoiceBench, KOpenAudioBench, and KMMAU using agent-driven transfer frameworks from English benchmarks and Korean ASR data, then evaluates eight SpeechLMs to show model-specific gaps and complementary weaknesses between SpokenQA and audio understanding.
Self-evolving rubric with anti-gaming fitness reveals that objective capability scaling fails to transfer to subjective LLM behaviors, with advice-restraint as the universal lowest dimension that can regress.
EduVideoBench is a new KSA-grounded benchmark that evaluates five frontier video generation models and finds substantial gaps in educational validity across knowledge, skills, and attitudes.
ContextEcho benchmark shows persona drift occurs across 23 frontier models in long agentic-coding sessions, is not reliably reset by compaction, and can be restored by single-shot anchors with mode-dependent effects.
GS-QA is a new benchmark of 2,800 QA pairs on 28 templates using OSM and Wikipedia data to evaluate LLMs on spatial predicates, multi-source reasoning, and diverse answer types including distances and counts.
Systematic review of thirteen malicious-code prompt corpora for coding LLM refusal evaluation that catalogs construction methods, surfaces gaps in human baselines, cross-corpus comparability, and malware taxonomies, and proposes methodological improvements.
ConsumerSimBench evaluates 13 LLMs on reconstructing crowd reactions from 1,553 Chinese social-media topics using 23,122 auditable yes-no criteria, finding maximum coverage of 47.8% by Gemini-3.1-Pro.
CBEA with LCV bounds evidence sets and validates commitments before response generation, achieving zero failures in scoped tests at 0.49-0.60 availability versus near-zero for baselines.
LLM multi-agent systems on lattices show bias-driven order-disorder crossovers instead of true phase transitions, with extracted effective couplings and fields serving as model-specific fingerprints.
citing papers explorer
-
FollowTable: A Benchmark for Instruction-Following Table Retrieval
FollowTable is the first large-scale benchmark for instruction-following table retrieval, paired with an Instruction Responsiveness Score, showing that existing models fail to adapt to fine-grained constraints beyond topical similarity.
-
MathNet: a Global Multimodal Benchmark for Mathematical Reasoning and Retrieval
MathNet delivers the largest multilingual Olympiad math dataset and benchmarks where models like Gemini-3.1-Pro reach 78% on solving but embedding models struggle on equivalent problem retrieval, with retrieval augmentation yielding up to 12% gains.
-
GIANTS: Generative Insight Anticipation from Scientific Literature
GIANTS-4B, trained with RL on a new 17k-example benchmark of parent-to-child paper insights, achieves 34% relative improvement over gemini-3-pro in LM-judge similarity and is rated higher-impact by a citation predictor.
-
FARS: A Fully Automated Research System Deployed at Scale
FARS deployed at scale produced 166 AI/ML papers across 67 topics that received 282 structured human reviews indicating some review-worthy outputs alongside recurring failure modes.
-
COHORT: Collaborative Orchestration for Hardening via Offensive Replay on Emulated Topologies
COHORT automates mitigation generation for network attacks via collaborative LLMs on emulated topologies with offensive replay evaluation, reporting 46.7% success rate that is 4.4 times higher than a single-agent baseline.
-
LLM-Based Examination of Eligibility Criteria from Securities Prospectuses at the German Central Bank
LLMs are applied in a generative pipeline for extracting, normalizing, and interpreting eligibility criteria from securities prospectuses, achieving up to 91% precision in document-level decisions with a conservative bias.
-
RealMath-Eval: Why SOTA Judges Struggle with Real Human Reasoning
RealMath-Eval benchmark shows LLM judges have an evaluation gap, performing worse on diverse real human math reasoning than on synthetic solutions due to greater error diversity and higher surprisal.
-
ReasoningFlow: Discourse Structures for Understanding LLM Reasoning Traces
ReasoningFlow represents LLM reasoning traces as DAGs, finding structural similarity across models and that most erroneous steps are unused in final answers.
-
CoEval: Ranking Language Models for Custom Tasks Without Labeled Data or Trustworthy Benchmarks
CoEval generates task-specific benchmarks by rotating models through teacher, student, and judge roles, then weights questions by discriminative power and judges by panel consensus to recover accurate model rankings without labels.
-
Synthetic Hallucinations, Real Gains: Hard Negatives from Frontier Models for FIM Hallucination Mitigation
Using frontier models to synthesize plausible-but-wrong FIM completions as hard negatives for SFT improves Delulu exact match by +18.8 and edit similarity by +0.22 on Qwen2.5-Coder-7B while also lifting HumanEval-Infilling and SAFIM.
-
Matter to Mechanism: A Benchmark for AI Co-Scientists in Materials and Battery Research
Introduces the Matter to Mechanism benchmark of 2,645 structured instances and a composite metric suite for evaluating AI co-scientists on problem-to-hypothesis reasoning in battery materials research.
-
RWGBench: Evaluating Scholarly Positioning in Related Work Generation
RWGBench is a citation-centric benchmark for related work generation built from 40k CS papers and a 100-paper test set, with multi-dimensional metrics that better match human expert judgment than standard similarity scores.
-
A Visually Impaired Assistance Benchmark for VLM-as-a-Judge Evaluation
VIABLE benchmark reveals existing VLM judges are unreliable for VIA tasks (GPT-5.4 at 52.6% diagnostic accuracy with 94.2% self-preference) and proposes VIA-Judge-Agent for improvements.
-
DirectorBench: Diagnosing Long-Form Video Generation with Personalized Multi-Agent Evaluation
DirectorBench is a profile-aware diagnostic benchmark that localizes bottlenecks in long-form video generation workflows using structured checkpoints and multi-agent evaluation.
-
KVoiceBench, KOpenAudioBench, and KMMAU: Agent-Driven Korean Speech Benchmarks for Evaluating SpeechLMs
Constructs KVoiceBench, KOpenAudioBench, and KMMAU using agent-driven transfer frameworks from English benchmarks and Korean ASR data, then evaluates eight SpeechLMs to show model-specific gaps and complementary weaknesses between SpokenQA and audio understanding.
-
Does Capability Transfer to Subjective Behavior -- and Would Our Instruments Tell Us? A Self-Evolving, Trust-by-Construction Evaluation Paradigm
Self-evolving rubric with anti-gaming fitness reveals that objective capability scaling fails to transfer to subjective LLM behaviors, with advice-restraint as the universal lowest dimension that can regress.
-
Are Video Models Zero-Shot Learners and Reasoners in Education? EduVideoBench, A Knowledge-Skills-Attitude Benchmark for Educational Video Generation
EduVideoBench is a new KSA-grounded benchmark that evaluates five frontier video generation models and finds substantial gaps in educational validity across knowledge, skills, and attitudes.
-
ContextEcho: A Benchmark for Persona Drift in Long Agentic-Coding Sessions
ContextEcho benchmark shows persona drift occurs across 23 frontier models in long agentic-coding sessions, is not reliably reset by compaction, and can be restored by single-shot anchors with mode-dependent effects.
-
GS-QA: A Benchmark for Geospatial Question Answering
GS-QA is a new benchmark of 2,800 QA pairs on 28 templates using OSM and Wikipedia data to evaluate LLMs on spatial predicates, multi-source reasoning, and diverse answer types including distances and counts.
-
Refusal Evaluation in Coding LLMs and Code Agents: A Systematic Review of Thirteen Malicious-Code Prompt Corpora (2023-2025)
Systematic review of thirteen malicious-code prompt corpora for coding LLM refusal evaluation that catalogs construction methods, surfaces gaps in human baselines, cross-corpus comparability, and malware taxonomies, and proposes methodological improvements.
-
Can LLMs Think Like Consumers? Benchmarking Crowd-Level Reaction Reconstruction with ConsumerSimBench
ConsumerSimBench evaluates 13 LLMs on reconstructing crowd reactions from 1,553 Chinese social-media topics using 23,122 auditable yes-no criteria, finding maximum coverage of 47.8% by Gemini-3.1-Pro.
-
Recall Isn't Enough: Bounding Commitments in Personalized Language Systems
CBEA with LCV bounds evidence sets and validates commitments before response generation, achieving zero failures in scoped tests at 0.49-0.60 availability versus near-zero for baselines.
-
Collective Alignment in LLM Multi-Agent Systems: Disentangling Bias from Cooperation via Statistical Physics
LLM multi-agent systems on lattices show bias-driven order-disorder crossovers instead of true phase transitions, with extracted effective couplings and fields serving as model-specific fingerprints.
-
StereoTales: A Multilingual Framework for Open-Ended Stereotype Discovery in LLMs
StereoTales shows that all tested LLMs emit harmful stereotypes in open-ended stories, with associations adapting to prompt language and targeting locally salient groups rather than transferring uniformly across languages.
-
Task-Aware Calibration: Provably Optimal Decoding in LLMs
Task calibration aligns LLM distributions in latent task spaces to make MBR decoding provably optimal and improve generation quality.
-
BIM Information Extraction Through LLM-based Adaptive Exploration
LLM adaptive exploration via runtime code execution outperforms static query generation for information extraction from heterogeneous BIM models on the new ifc-bench v2 benchmark.
-
ClassEval-Pro: A Cross-Domain Benchmark for Class-Level Code Generation
ClassEval-Pro benchmark shows frontier LLMs achieve at most 45.6% Pass@1 on class-level code tasks, with logic errors (56%) and dependency errors (38%) as dominant failure modes.
-
PSI-Bench: Towards Clinically Grounded and Interpretable Evaluation of Depression Patient Simulators
Depression patient simulators produce overly long, low-variability responses that resolve emotions too quickly along a uniform trajectory, with framework choice outweighing model scale.
-
Beyond Accuracy: Benchmarking Cross-Task Consistency in Unified Multimodal Models
XTC-Bench reveals that strong performance on generation or understanding tasks in unified multimodal models does not guarantee cross-task semantic consistency, which instead depends on how tightly coupled the learning objectives are across modalities.
-
Navigating Large-Scale Document Collections: MuDABench for Multi-Document Analytical QA
MuDABench provides 332 analytical QA instances over large semi-structured document collections, showing standard RAG performs poorly while a multi-agent workflow with planning, extraction, and code generation improves results but leaves a gap to human experts.
-
Semantic Needles in Document Haystacks: Sensitivity Testing of LLM-as-a-Judge Similarity Scoring
LLMs exhibit positional bias and context-dependent scoring patterns when judging document similarity, with each model showing a stable scoring fingerprint but a shared hierarchy of sensitivity to different semantic perturbations.
-
MM-JudgeBias: A Benchmark for Evaluating Compositional Biases in MLLM-as-a-Judge
MM-JudgeBias benchmark shows that many MLLM judges neglect modalities and produce unstable evaluations under small input changes, based on tests of 26 models with over 1,800 samples.
-
Evaluating Tool-Using Language Agents: Judge Reliability, Propagation Cascades, and Runtime Mitigation in AgentProp-Bench
AgentProp-Bench shows substring judging agrees with humans at kappa=0.049, LLM ensemble at 0.432, bad-parameter injection propagates with ~0.62 probability, rejection and recovery are independent, and a runtime fix cuts hallucinations 23pp on GPT-4o-mini but not Gemini-2.0-Flash.
-
LLM-Based Data Generation and Clinical Skills Evaluation for Low-Resource French OSCEs
A controlled LLM pipeline generates synthetic French OSCE transcripts with varying skill levels and evaluates them, with mid-size models achieving ~90% accuracy matching GPT-4o on the synthetic data.
-
Generate, Filter, Control, Replay: A Comprehensive Survey of Rollout Strategies for LLM Reinforcement Learning
This survey introduces the Generate-Filter-Control-Replay (GFCR) taxonomy to structure rollout pipelines for RL-based post-training of reasoning LLMs.
-
PR-CAD: Progressive Refinement for Unified Controllable and Faithful Text-to-CAD Generation with Large Language Models
PR-CAD unifies text-to-CAD generation and editing via progressive refinement with LLMs, a new interaction dataset, and RL-enhanced reasoning to achieve better controllability and faithfulness.
-
When Negation Is a Geometry Problem in Vision-Language Models
A direction associated with negation exists in CLIP embedding space and can be steered at test time via representation engineering to produce negation-aware outputs without fine-tuning.
-
Using LLM-as-a-Judge/Jury to Advance Scalable, Clinically-Validated Safety Evaluations of Model Responses to Users Demonstrating Psychosis
Seven clinician-informed safety criteria enable LLM-as-a-Judge to reach substantial agreement with human consensus (Cohen's κ up to 0.75) on evaluating LLM responses to users demonstrating psychosis.
-
Beyond Verifiable Rewards: Rubric-Based GRM for Reinforced Fine-Tuning SWE Agents
A rubric-based generative reward model improves reinforced fine-tuning of SWE agents by supplying richer behavioral guidance than binary terminal rewards alone.
-
When Agents Fail: A Comprehensive Study of Bugs in LLM Agents with Automated Labeling
A large-scale empirical study categorizes bugs in LLM agents and demonstrates that a specialized LLM agent can annotate them accurately at very low cost.
-
Membership Inference Attacks for Retrieval Based In-Context Learning for Document Question Answering
Black-box membership inference attacks on retrieval-based in-context learning for document QA succeed via query prefixes, with a novel weighted-averaging method outperforming priors even under paraphrasing.
-
VIDA: A dataset for Visually Dependent Ambiguity in Multimodal Machine Translation
VIDA provides 2,500 visually-dependent ambiguous translation examples and span-level disambiguation metrics; CoT-SFT on LVLMs improves out-of-distribution performance over standard SFT.
-
Theoria: Rewrite-Acceptability Verification over Informal Reasoning States
Theoria rewrites solutions into auditable typed state transitions with justifications, certifying 105 of 185 HLE problems at 91.4% precision and outperforming holistic judges on adversarial poisoned proofs by catching hidden premises.
-
Rigel: Self-Distilled Score Adaptation for Image and Video Captioning Evaluation
Rigel is a self-distilled LLM-based metric for image and video caption evaluation that reports over 10-point gains on ActivityNet-Fact in reference-free settings.
-
Stop Hand-Holding Your Coding Agent: Engineering the Loops that Replace Step-by-Step Prompting
Introduces loop engineering as a distinct practice layer for coding agents, supplies a taxonomy and verification ladder, and analyzes a hand-coded corpus of fifty real loops.
-
HistoRAG: Embedding Historical Methodology in Retrieval-Augmented Generation Through Critical Technical Practice
HistoRAG embeds historiographical principles into RAG via temporal windowing, decoupled retrieval, and contestable LLM relevance judgments, evaluated on 102k Der Spiegel articles from 1950-1979.
-
Catching One in Five: LLM-as-Judge Blind Spots in Production Multi-Turn Transaction Agents
Empirical study of a production multi-turn ordering agent finds LLM-as-judge recall below 25% for human-confirmed defects, missing cross-turn state issues due to limited rubric and routing.
-
LLM-Orchestrated Conformance Checking in Stroke Care Without Computer-Interpretable Guidelines
An LLM-orchestrated framework enables conformance checking in stroke care from unstructured texts, achieving over 86% conformance in hospital data.
-
Beyond Scalar Rewards by Internalizing Reasoning into Score Distributions
Z-Reward trains a 27B reasoning teacher VLM on score distributions via GDSO and distills it via RISD into a 9B student, reaching 89.6% and 88.6% human preference accuracy with 41.3% optimization gain over SFT baseline.
-
Learn from Your Mistakes: Tree-like Self-Play for Secure Code LLMs
TSP reframes secure code generation as a tree-structured self-play process that supplies dense on-policy signals at vulnerability-prone nodes, yielding higher security pass rates and cross-language generalization than SFT or unstructured self-play.