探针论文趋势 2023–2026

主要结论

• 探针论文从 2023 年的 10 篇增长到 2026 年的 15 篇，三年间扩大到约 1.5 倍。
• NeurIPS 是发探针论文最多的会议，2023–2026 年共 12 篇，占四个会议总量的 28%。
• 在「可解释性」的 5 个主题里，它的 2026 年论文数排第 5；同组增长最快的机制解释是 8.1 倍，探针是 1.5 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 2026Optimal Representation Size: High-Dimensional Analysis of Pretraining and Linear Probing
• ICML 2026Logit Distance Bounds Representational Similarity
• ICLR 2026Probing Rotary Position Embeddings through Frequency Entropy
• CVPR 2026Do VLMs Perceive or Recall Probing Visual Perception vs. Memory
• NeurIPS 2025SparseMVC: Probing Cross-view Sparsity Variations for Multi-view Clustering
• ICML 2025ViTally Consistent: Scaling Biological Representation Learning for Cell Microscopy
• ICLR 2025Breaking Free from MMI: A New Frontier in Rationalization by Probing Input Utilization
• CVPR 2025MetricGrids: Arbitrary Nonlinear Approximation with Elementary Metric Grids based Implicit Neural Representation
• NeurIPS 2024EMVP: Embracing Visual Foundation Model for Visual Place Recognition with Centroid-Free Probing
• ICML 2024Position: The Platonic Representation Hypothesis
• ICLR 2024LiDAR: Sensing Linear Probing Performance in Joint Embedding SSL Architectures
• CVPR 2024Improved Implicit Neural Representation with Fourier Reparameterized Training
• ICML 2023Defects of Convolutional Decoder Networks in Frequency Representation
• ICLR 2023Can BERT Refrain from Forgetting on Sequential Tasks? A Probing Study
• CVPR 2023Probing Neural Representations of Scene Perception in a Hippocampally Dependent Task Using Artificial Neural Networks
• NeurIPS 2026Towards Precise Knowledge Distillation for Large Language Models via Knowledge Probing
• ICML 2026Probing the Inductive Bias of Neural Networks through Learning Random Cellular Automata
• ICLR 2026Rethinking LLM-as-a-Judge: Representation-as-a-Judge with Small Language Models via Semantic Capacity Asymmetry
• CVPR 2026Faithful Contouring Near-Lossless 3D Voxel Representation Free from Iso-surface
• NeurIPS 2025MaxSup: Overcoming Representation Collapse in Label Smoothing

全部探针论文

共 15 篇，第 1 / 1 页

1. CVPR 2026Do VLMs Perceive or Recall Probing Visual Perception vs. Memory
2. CVPR 2026Faithful Contouring Near-Lossless 3D Voxel Representation Free from Iso-surface
3. ICLR 2026Attention, Please! Revisiting Attentive Probing Through the Lens of Efficiency
4. ICLR 2026Probing Rotary Position Embeddings through Frequency Entropy
5. ICLR 2026Rethinking LLM-as-a-Judge: Representation-as-a-Judge with Small Language Models via Semantic Capacity Asymmetry
6. ICML 2026Logit Distance Bounds Representational Similarity
7. ICML 2026Probing the Geometry of Diffusion Models with the String Method
8. ICML 2026Probing the Inductive Bias of Neural Networks through Learning Random Cellular Automata
9. ICML 2026The Perception–Physics Paradox: Probing Scientific Alignment with TC-Bench
10. NeurIPS 2026Is a Linear Probe Evidence of a Linear Representation?
11. NeurIPS 2026Optimal Representation Size: High-Dimensional Analysis of Pretraining and Linear Probing
12. NeurIPS 2026Probing the Trajectories of Reasoning Traces in Large Language Models
13. NeurIPS 2026Prototypes of the Mind: A Unified Framework for Probing the Visual Brain
14. NeurIPS 2026Towards Precise Knowledge Distillation for Large Language Models via Knowledge Probing
15. NeurIPS 2026Value Entanglement: Conflation Between Different Kinds of Good In (Some) Large Language Models

常见问题

2023–2026 年四大会议一共有多少篇探针论文？

共 43 篇，按年份分别是 2023 年 10 篇、2024 年 5 篇、2025 年 13 篇、2026 年 15 篇。

NeurIPS 2026 有多少篇探针论文？

NeurIPS 2026 有 6 篇探针论文，上一届是 4 篇。

ICLR 2026 有多少篇探针论文？

ICLR 2026 有 3 篇探针论文，上一届是 4 篇。

ICML 2026 有多少篇探针论文？

ICML 2026 有 4 篇探针论文，上一届是 1 篇。

CVPR 2026 有多少篇探针论文？

CVPR 2026 有 2 篇探针论文，上一届是 4 篇。

哪个会议的探针论文最多？

NeurIPS，2023–2026 年共 12 篇。