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 126 篇，第 2 / 3 页

51. ICML 2025Emoji Attack: Enhancing Jailbreak Attacks Against Judge LLM Detection
52. ICML 2025Evaluating Judges as Evaluators: The JETTS Benchmark of LLM-as-Judges as Test-Time Scaling Evaluators
53. ICML 2025Exploring Criteria of Loss Reweighting to Enhance LLM Unlearning
54. ICML 2025From Jack of All Trades to Master of One: Specializing LLM-based Autoraters to a Test Set
55. ICML 2025Great Models Think Alike and this Undermines AI Oversight
56. ICML 2025Investigating Non-Transitivity in LLM-as-a-Judge
57. ICML 2025La RoSA: Enhancing LLM Efficiency via Layerwise Rotated Sparse Activation
58. ICML 2025Large Language-Geometry Model: When LLM meets Equivariance
59. ICML 2025Learning to Plan & Reason for Evaluation with Thinking-LLM-as-a-Judge
60. ICML 2025MAGELLAN: Metacognitive predictions of learning progress guide autotelic LLM agents in large goal spaces
61. ICML 2025Memorization Sinks: Isolating Memorization during LLM Training
62. ICML 2025Monte Carlo Tree Search for Comprehensive Exploration in LLM-Based Automatic Heuristic Design
63. ICML 2025Overestimation in LLM Evaluation: A Controlled Large-Scale Study on Data Contamination’s Impact on Machine Translation
64. ICML 2025Position: Don't Use the CLT in LLM Evals With Fewer Than a Few Hundred Datapoints
65. ICML 2025Position: LLM Social Simulations Are a Promising Research Method
66. ICML 2025Position: Principles of Animal Cognition to Improve LLM Evaluations
67. ICML 2025Position: The Most Expensive Part of an LLM *should* be its Training Data
68. ICML 2025Prompt-to-Leaderboard: Prompt-Adaptive LLM Evaluations
69. ICML 2025RZ-NAS: Enhancing LLM-guided Neural Architecture Search via Reflective Zero-Cost Strategy
70. ICML 2025The Emperor's New Clothes in Benchmarking? A Rigorous Examination of Mitigation Strategies for LLM Benchmark Data Contamination
71. ICML 2025Tuning LLM Judge Design Decisions for 1/1000 of the Cost
72. ICML 2025TypyBench: Evaluating LLM Type Inference for Untyped Python Repositories
73. ICML 2025Uncertainty Quantification for LLM-Based Survey Simulations
74. NeurIPS 2025$\texttt{BetaConform}$: Efficient MAP Estimation of LLM Ensemble Judgment Performance with Prior Transfer
75. NeurIPS 2025ACT as Human: Multimodal Large Language Model Data Annotation with Critical Thinking
76. NeurIPS 2025Adaptive Prediction-Powered AutoEval with Reliability and Efficiency Guarantees
77. NeurIPS 2025Any Large Language Model Can Be a Reliable Judge: Debiasing with a Reasoning-based Bias Detector
78. NeurIPS 2025Ask a Strong LLM Judge when Your Reward Model is Uncertain
79. NeurIPS 2025AutoJudge: Judge Decoding Without Manual Annotation
80. NeurIPS 2025Beyond the Surface: Enhancing LLM-as-a-Judge Alignment with Human via Internal Representations
81. NeurIPS 2025Bohdi: Heterogeneous LLM Fusion with Automatic Data Exploration
82. NeurIPS 2025BrainEC-LLM: Brain Effective Connectivity Estimation by Multiscale Mixing LLM
83. NeurIPS 2025Bridging Human and LLM Judgments: Understanding and Narrowing the Gap
84. NeurIPS 2025Comparison requires valid measurement: Rethinking attack success rate comparisons in AI red teaming
85. NeurIPS 2025Correlation Dimension of Autoregressive Large Language Models
86. NeurIPS 2025Cost-aware LLM-based Online Dataset Annotation
87. NeurIPS 2025Distributional LLM-as-a-Judge
88. NeurIPS 2025Do Automatic Factuality Metrics Measure Factuality? A Critical Evaluation
89. NeurIPS 2025Do Language Models Use Their Depth Efficiently?
90. NeurIPS 2025Enhancing Multilingual LLM Pretraining with Model-Based Data Selection
91. NeurIPS 2025Evaluating LLM-contaminated Crowdsourcing Data Without Ground Truth
92. NeurIPS 2025Flex-Judge: Text-Only Reasoning Unleashes Zero-Shot Multimodal Evaluators
93. NeurIPS 2025From Replication to Redesign: Exploring Pairwise Comparisons for LLM-Based Peer Review
94. NeurIPS 2025GPAS: Accelerating Convergence of LLM Pretraining via Gradient-Preserving Activation Scaling
95. NeurIPS 2025HyGen: Efficient LLM Serving via Elastic Online-Offline Request Co-location
96. NeurIPS 2025KORGym: A Dynamic Game Platform for LLM Reasoning Evaluation
97. NeurIPS 2025Less is More: Improving LLM Alignment via Preference Data Selection
98. NeurIPS 2025Let the LLM Stick to Its Strengths: Learning to Route Economical LLM
99. NeurIPS 2025LiveCodeBench Pro: How Do Olympiad Medalists Judge LLMs in Competitive Programming?
100. NeurIPS 2025Lua-LLM: Learning Unstructured-Sparsity Allocation for Large Language Models

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。