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 108 篇，第 1 / 3 页

1. ICML 2026A Coin Flip for Safety: LLM Judges Fail to Reliably Measure Adversarial Robustness
2. ICML 2026A Judge-Aware Ranking Framework for Evaluating Large Language Models without Ground Truth
3. ICML 2026Adaptive Contracts for Cost-Effective AI Delegation
4. ICML 2026Are LLM Evaluators Really Narcissists? Sanity Checking Self-Preference Evaluations
5. ICML 2026Automatic Construction of Clinical Scoring Systems with LLM Agents
6. ICML 2026Beyond Drift: Stabilizing Subjective LLM Evaluation with Information-Theoretic Rubrics
7. ICML 2026Breaking the Exploration Bottleneck: Rubric-Scaffolded Reinforcement Learning for Open-Ended LLM Reasoning
8. ICML 2026CARE: Confounder-Aware Aggregation for Reliable LLM Evaluation
9. ICML 2026Controlled LLM Training on Spectral Sphere
10. ICML 2026Cutting LLM Evaluation Costs with SySRs: A Bandit Algorithm That Provably Exploits Model Similarity
11. ICML 2026D-Judge: Disrupting Multi-Turn Jailbreaks using Semantics-Preserving Output Rewriting
12. ICML 2026Demystifying LLM-as-a-Judge: Analytically Tractable Model for Inference-Time Scaling
13. ICML 2026Dependence-Aware Label Aggregation for LLM-as-a-Judge via Ising Models
14. ICML 2026Diagnosing the Reliability of LLM-as-a-Judge via Item Response Theory
15. ICML 2026Distribution-Calibrated Inference Time Compute for Thinking LLM-as-a-Judge
16. ICML 2026Distributional Open-Ended Evaluation of LLM Cultural Value Alignment Based on Value Codebook
17. ICML 2026Does Reinforcement Fine-Tuning Improve Generalization of LLM Agents? An Empirical Study
18. ICML 2026Easier to Judge than to Find: Predicting In-Context Learning Success for Demonstration Selection
19. ICML 2026Efficient Distributed MLLM Training with Cornstarch
20. ICML 2026Efficient Inference for Noisy LLM-as-a-Judge Evaluation
21. ICML 2026Enhancing LLM Training via Spectral Clipping
22. ICML 2026Evaluating Language Models in Realistic Conversational Contexts
23. ICML 2026ExCyTIn-Bench: Evaluating LLM agents on Cyber Threat Investigation
24. ICML 2026FairJudge : An Adaptive, Debiased, and Consistent LLM-as-a-Judge
25. ICML 2026FiRE: Fine-grained Ranking Evaluation for Machine Translation
26. ICML 2026Game-Theoretic Co-Evolution for LLM-Based Heuristic Discovery
27. ICML 2026How to Correctly Report LLM-as-a-Judge Evaluations
28. ICML 2026InnoEval: On Research Idea Evaluation as a Knowledge-Grounded, Multi-Perspective Reasoning Problem
29. ICML 2026JADE: Expert-Grounded Dynamic Evaluation for Open-Ended Professional Tasks
30. ICML 2026Judging What We Cannot Solve: A Consequence-Based Approach for Oracle-Free Evaluation of Research-Level Math
31. ICML 2026Learn-to-learn on Arbitrary Textual Conditioning: A Hypernetwork-Driven Meta-gated LLM
32. ICML 2026LitReview Arena: Evaluating Literature Review Agents with Battle-style Peer Review Platform
33. ICML 2026LLM Priors for ERM over Programs
34. ICML 2026LLM Watermark Evasion via Bias Inversion
35. ICML 2026LLM-based Embeddings: Attention Values Encode Sentence Semantics Better Than Hidden States
36. ICML 2026Margin-Adaptive Confidence Ranking for Reliable LLM Judgement
37. ICML 2026Mitigating Perceptual Judgment Bias in Multimodal LLM-as-a-Judge via Perceptual Perturbation and Reward Modeling
38. ICML 2026Mitigating Translationese Bias in Multilingual LLM-as-a-Judge via Disentangled Information Bottleneck
39. ICML 2026No Free Lunch: Non-Asymptotic Analysis of Prediction-Powered Inference
40. ICML 2026Nonparametric LLM Evaluation from Preference Data
41. ICML 2026On the Limits of LLM Adaptability: Impact of Model-Internalized Priors on Annotation Task Performance
42. ICML 2026Order Matters: Unveiling the Hidden Impact of Macro Placement Sequences via Proxy-Guided LLM Evolution
43. ICML 2026PPDL: LLM-Based Flows as Probabilistic Programs
44. ICML 2026PPT-Eval: A Benchmark for Computer-Use Agents on PowerPoint Tasks
45. ICML 2026Prediction-Powered Risk Monitoring of Deployed Models for Detecting Harmful Distribution Shifts
46. ICML 2026Pressure Reveals Character: Behavioural Alignment Evaluation at Depth
47. ICML 2026ProbeLLM: Automating Principled Diagnosis of LLM Failures
48. ICML 2026ProtocolBench: Which LLM MultiAgent Protocol to Choose?
49. ICML 2026Quantifying Biases in LLM-as-a-Judge Evaluations
50. ICML 2026Quartet II: Accurate LLM Pre-Training in NVFP4 by Improved Unbiased Gradient Estimation

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。