模型即裁判论文趋势 2023–2026

主要结论

• 模型即裁判论文从 2023 年的 8 篇增长到 2026 年的 219 篇，三年间扩大到约 27 倍。
• NeurIPS 是发模型即裁判论文最多的会议，2023–2026 年共 173 篇，占四个会议总量的 43%。
• 交叉方向里涨得最快的是AI 智能体：同时属于两个主题的论文从 2023 年的 1 篇增至 2026 年的 78 篇。
• 在「评测与基准」的 5 个主题里，它的 2026 年论文数排第 2；同组增长最快的大模型基准是 154 倍，模型即裁判是 27 倍。

代表论文

按年份和会议均匀挑选，方便快速看到每一届的样貌

• NeurIPS 2026A Free Lunch in LLM Compression: Revisiting Retraining after Pruning
• ICML 2026A Judge-Aware Ranking Framework for Evaluating Large Language Models without Ground Truth
• ICLR 2026Addressing Pitfalls in the Evaluation of Uncertainty Estimation Methods for Natural Language Generation
• CVPR 2026Pixels Dont Lie But Your Detector Might Bootstrapping MLLM-as-a-Judge for
• NeurIPS 2025$\texttt{BetaConform}$: Efficient MAP Estimation of LLM Ensemble Judgment Performance with Prior Transfer
• ICML 2025Active Evaluation Acquisition for Efficient LLM Benchmarking
• ICLR 2025An Empirical Analysis of Uncertainty in Large Language Model Evaluations
• CVPR 2025ASHiTA: Automatic Scene-grounded HIerarchical Task Analysis
• NeurIPS 2024BertaQA: How Much Do Language Models Know About Local Culture?
• ICML 2024CHAI: Clustered Head Attention for Efficient LLM Inference
• ICLR 2024PandaLM: An Automatic Evaluation Benchmark for LLM Instruction Tuning Optimization
• NeurIPS 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena
• ICML 2023SAM operates far from home: eigenvalue regularization as a dynamical phenomenon
• ICLR 2023On the complexity of nonsmooth automatic differentiation
• NeurIPS 2026A Stratified Multi-Rater Evaluation of LLM-Based Virtual Standardized Patients with a Deployed Data-Generation Platform
• ICML 2026Are LLM Evaluators Really Narcissists? Sanity Checking Self-Preference Evaluations
• ICLR 2026AutoMetrics: Approximate Human Judgments with Automatically Generated Evaluators
• CVPR 20263DrawAgent Teaching LLM to Draw in 3D with Early Contrastive
• NeurIPS 2025Adaptive Prediction-Powered AutoEval with Reliability and Efficiency Guarantees
• ICML 2025Evaluating Judges as Evaluators: The JETTS Benchmark of LLM-as-Judges as Test-Time Scaling Evaluators

全部模型即裁判论文

共 62 篇，第 1 / 2 页

1. ICLR 2026Addressing Pitfalls in the Evaluation of Uncertainty Estimation Methods for Natural Language Generation
2. ICLR 2026Align Once, Benefit Multilingually: Enforcing Multilingual Consistency for LLM Safety Alignment
3. ICLR 2026Align to Misalign: Automatic LLM Jailbreak with Meta-Optimized LLM Judges
4. ICLR 2026Auto-RT: Automatic Jailbreak Strategy Exploration for Red-Teaming Large Language Models
5. ICLR 2026AutoCodeBench: Large Language Models are Automatic Code Benchmark Generators
6. ICLR 2026AutoLibra: Agent Metric Induction from Open-Ended Human Feedback
7. ICLR 2026AutoMetrics: Approximate Human Judgments with Automatically Generated Evaluators
8. ICLR 2026AutoTool: Automatic Scaling of Tool-Use Capabilities in RL via Decoupled Entropy Constraints
9. ICLR 2026Benchmarking LLM Tool-Use in the Wild
10. ICLR 2026BiasScope: Towards Automated Detection of Bias in LLM-as-a-Judge Evaluation
11. ICLR 2026Counterfactual LLM-based Framework for Measuring Rhetorical Style
12. ICLR 2026DiscoX: Benchmarking Discourse-Level Translation in Expert Domains
13. ICLR 2026DiSRouter: Distributed Self-Routing for LLM Selections
14. ICLR 2026Doubly-Robust LLM-as-a-Judge: Externally Valid Estimation with Imperfect Personas
15. ICLR 2026DUET: Optimizing LLM Training Data Mixtures via Noisy Feedback from Unseen, Downstream Evaluation Tasks
16. ICLR 2026Epistemic Uncertainty Quantification To Improve Decisions From Black-Box Models
17. ICLR 2026Fewer Battles, More Gain: An Information-Efficient Framework for Arena-based LLM Evaluation
18. ICLR 2026Fewer Weights, More Problems: A Practical Attack on LLM Pruning
19. ICLR 2026Foundational Automatic Evaluators: Scaling Multi-Task Generative Evaluator Training for Reasoning-Centric Domains
20. ICLR 2026Generative Value Conflicts Reveal LLM Priorities
21. ICLR 2026HiFo-Prompt: Prompting with Hindsight and Foresight for LLM-based Automatic Heuristic Design
22. ICLR 2026How Learning Rate Decay Wastes Your Best Data in Curriculum-Based LLM Pretraining
23. ICLR 2026Incentivizing Agentic Reasoning in LLM Judges via Tool-Integrated Reinforcement Learning
24. ICLR 2026Influence-Preserving Proxies for Gradient-Based Data Selection in LLM FineTuning
25. ICLR 2026J1: Incentivizing Thinking in LLM-as-a-Judge via Reinforcement Learning
26. ICLR 2026LLM Fingerprinting via Semantically Conditioned Watermarks
27. ICLR 2026LLM-as-a-Prophet: Understanding Predictive Intelligence with Prophet Arena
28. ICLR 2026LLM-JEPA: Large Language Models Meet Joint Embedding Predictive Architectures
29. ICLR 2026LLMS ON TRIAL: Evaluating Judicial Fairness For Large Language Models
30. ICLR 2026MENLO: From Preferences to Proficiency – Evaluating and Modeling Native-like Quality Across 47 Languages
31. ICLR 2026Meta-Router: Bridging Gold-standard and Preference-based Evaluations in LLM Routing
32. ICLR 2026Multiple-Prediction-Powered Inference
33. ICLR 2026Noisy but Valid: Robust Statistical Evaluation of LLMs with Imperfect Judges
34. ICLR 2026On the Shelf Life of Finetuned LLM-Judges: Future Proofing, Backward Compatibility, and Question Generalization
35. ICLR 2026Predicting LLM Output Length via Entropy-Guided Representations
36. ICLR 2026Predicting LLM Reasoning Performance with Small Proxy Model
37. ICLR 2026Predicting Training Re-evaluation Curves Enables Effective Data Curriculums for LLMs
38. ICLR 2026Preference Leakage: A Contamination Problem in LLM-as-a-judge
39. ICLR 2026ProfBench: Multi-Domain Rubrics requiring Professional Knowledge to Answer and Judge
40. ICLR 2026ProRe: A Proactive Reward System for GUI Agents via Reasoner–Actor Collaboration
41. ICLR 2026QuoKA: Query-Oriented KV Selection for Efficient LLM Prefill
42. ICLR 2026QuRL: Rubrics As Judge For Open-Ended Question Answering
43. ICLR 2026References Improve LLM Alignment in Non-Verifiable Domains
44. ICLR 2026Reliable Fine-Grained Evaluation of Natural Language Math Proofs
45. ICLR 2026Rethinking LLM Evaluation: Can We Evaluate LLMs with 200× Less Data?
46. ICLR 2026Rethinking LLM-as-a-Judge: Representation-as-a-Judge with Small Language Models via Semantic Capacity Asymmetry
47. ICLR 2026RouterArena: An Open Platform for Comprehensive Comparison of LLM Routers
48. ICLR 2026SCI-Verifier: Scientific Verifier with Thinking
49. ICLR 2026Semantic Voting: A Self-Evaluation-Free Approach for Efficient LLM Self-Improvement on Unverifiable Open-ended Tasks
50. ICLR 2026SK2Decompile: LLM-based Two-Phase Binary Decompilation from Skeleton to Skin

常见问题

2023–2026 年四大会议一共有多少篇模型即裁判论文？

共 401 篇，按年份分别是 2023 年 8 篇、2024 年 48 篇、2025 年 126 篇、2026 年 219 篇。

NeurIPS 2026 有多少篇模型即裁判论文？

NeurIPS 2026 有 88 篇模型即裁判论文，上一届是 53 篇。

ICLR 2026 有多少篇模型即裁判论文？

ICLR 2026 有 62 篇模型即裁判论文，上一届是 37 篇。

ICML 2026 有多少篇模型即裁判论文？

ICML 2026 有 64 篇模型即裁判论文，上一届是 33 篇。

CVPR 2026 有多少篇模型即裁判论文？

CVPR 2026 有 5 篇模型即裁判论文，上一届是 3 篇。

哪个会议的模型即裁判论文最多？

NeurIPS，2023–2026 年共 173 篇。