机制解释论文趋势 2023–2026

主要结论

• 机制解释论文从 2023 年的 17 篇增长到 2026 年的 138 篇，三年间扩大到约 8.1 倍。
• ICML 是发机制解释论文最多的会议，2023–2026 年共 119 篇，占四个会议总量的 38%。
• 交叉方向里涨得最快的是深度学习 · large language models：同时属于两个主题的论文从 2023 年的 1 篇增至 2026 年的 23 篇。
• 在「可解释性」的 5 个主题里，它的 2026 年论文数排第 2；同组增长最快的特征归因是 1.9 倍，机制解释是 8.1 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 Mechanistic Interpretability Study of an Astronomical Foundation Model
• ICML 2026Contribution Weights: A Geometrical Analysis of Self-Attention Transformers
• ICLR 2026Circuit Insights: Towards Interpretability Beyond Activations
• CVPR 2026CIGMA Causal Information-Gain Mechanistic Attribution of Attention Heads in Vision
• NeurIPS 2025Beyond Components: Singular Vector-Based Interpretability of Transformer Circuits
• ICML 2025Algorithm Development in Neural Networks: Insights from the Streaming Parity Task
• ICLR 2025Arithmetic Without Algorithms: Language Models Solve Math with a Bag of Heuristics
• CVPR 2025Dissecting and Mitigating Diffusion Bias via Mechanistic Interpretability
• NeurIPS 2024InterpBench: Semi-Synthetic Transformers for Evaluating Mechanistic Interpretability Techniques
• ICML 2024Mechanistic Neural Networks for Scientific Machine Learning
• ICLR 2024Fine-Tuning Enhances Existing Mechanisms: A Case Study on Entity Tracking
• CVPR 2024SI-MIL: Taming Deep MIL for Self-Interpretability in Gigapixel Histopathology
• NeurIPS 2023A Logic for Expressing Log-Precision Transformers
• ICML 2023Mechanistic Mode Connectivity
• ICLR 2023Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 Small
• NeurIPS 2026Discovering Mechanistic Models of Neural Activity: System Identification in an in Silico Zebrafish
• ICML 2026CorrSteer: Generation-Time LLM Steering via Correlated Sparse Autoencoder Features
• ICLR 2026CLUE: Conflict-guided Localization for LLM Unlearning Framework
• CVPR 2026From Weights to Concepts Data-Free Interpretability of CLIP via Singular
• NeurIPS 2025Bigram Subnetworks: Mapping to Next Tokens in Transformer Language Models

全部机制解释论文

共 33 篇，第 1 / 1 页

1. ICML 2025Algorithm Development in Neural Networks: Insights from the Streaming Parity Task
2. ICML 2025Are Sparse Autoencoders Useful? A Case Study in Sparse Probing
3. ICML 2025Beyond Induction Heads: In-Context Meta Learning Induces Multi-Phase Circuit Emergence
4. ICML 2025ConceptAttention: Diffusion Transformers Learn Highly Interpretable Features
5. ICML 2025Constrained Belief Updates Explain Geometric Structures in Transformer Representations
6. ICML 2025Emergence and Effectiveness of Task Vectors in In-Context Learning: An Encoder Decoder Perspective
7. ICML 2025Emergent Symbolic Mechanisms Support Abstract Reasoning in Large Language Models
8. ICML 2025Evaluating Neuron Explanations: A Unified Framework with Sanity Checks
9. ICML 2025Geometric Hyena Networks for Large-scale Equivariant Learning
10. ICML 2025How Do Transformers Learn Variable Binding in Symbolic Programs?
11. ICML 2025In-Context Linear Regression Demystified: Training Dynamics and Mechanistic Interpretability of Multi-Head Softmax Attention
12. ICML 2025Inference-Time Decomposition of Activations (ITDA): A Scalable Approach to Interpreting Large Language Models
13. ICML 2025Learning Multi-Level Features with Matryoshka Sparse Autoencoders
14. ICML 2025Locate-then-edit for Multi-hop Factual Recall under Knowledge Editing
15. ICML 2025Looking Beyond the Top-1: Transformers Determine Top Tokens in Order
16. ICML 2025Mechanistic Unlearning: Robust Knowledge Unlearning and Editing via Mechanistic Localization
17. ICML 2025MIB: A Mechanistic Interpretability Benchmark
18. ICML 2025Models of Heavy-Tailed Mechanistic Universality
19. ICML 2025Position: Explainable AI Cannot Advance Without Better User Studies
20. ICML 2025Position: Rethinking Explainable Machine Learning as Applied Statistics
21. ICML 2025Representation Shattering in Transformers: A Synthetic Study with Knowledge Editing
22. ICML 2025SAEBench: A Comprehensive Benchmark for Sparse Autoencoders in Language Model Interpretability
23. ICML 2025Scaling Sparse Feature Circuits For Studying In-Context Learning
24. ICML 2025Strategy Coopetition Explains the Emergence and Transience of In-Context Learning
25. ICML 2025Taming Knowledge Conflicts in Language Models
26. ICML 2025The Logical Implication Steering Method for Conditional Interventions on Transformer Generation
27. ICML 2025The underlying structures of self-attention: symmetry, directionality, and emergent dynamics in Transformer training
28. ICML 2025To Steer or Not to Steer? Mechanistic Error Reduction with Abstention for Language Models
29. ICML 2025Towards Global-level Mechanistic Interpretability: A Perspective of Modular Circuits of Large Language Models
30. ICML 2025Towards scientific discovery with dictionary learning: Extracting biological concepts from microscopy foundation models
31. ICML 2025Towards Understanding Fine-Tuning Mechanisms of LLMs via Circuit Analysis
32. ICML 2025Validating Mechanistic Interpretations: An Axiomatic Approach
33. ICML 2025Which Attention Heads Matter for In-Context Learning?

常见问题

2023–2026 年四大会议一共有多少篇机制解释论文？

共 311 篇，按年份分别是 2023 年 17 篇、2024 年 49 篇、2025 年 107 篇、2026 年 138 篇。

NeurIPS 2026 有多少篇机制解释论文？

NeurIPS 2026 有 16 篇机制解释论文，上一届是 34 篇。

ICLR 2026 有多少篇机制解释论文？

ICLR 2026 有 46 篇机制解释论文，上一届是 38 篇。

ICML 2026 有多少篇机制解释论文？

ICML 2026 有 74 篇机制解释论文，上一届是 33 篇。

CVPR 2026 有多少篇机制解释论文？

CVPR 2026 有 2 篇机制解释论文，上一届是 2 篇。

哪个会议的机制解释论文最多？

ICML，2023–2026 年共 119 篇。