推理加速论文趋势 2023–2026

主要结论

• 推理加速论文从 2023 年的 54 篇增长到 2026 年的 91 篇，三年间扩大到约 1.7 倍。
• NeurIPS 是发推理加速论文最多的会议，2023–2026 年共 90 篇，占四个会议总量的 35%。
• 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 9 篇。
• 在「效率与系统」的 5 个主题里，它的 2026 年论文数排第 5；同组增长最快的硬件协同是 4.4 倍，推理加速是 1.7 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 Method-Class Divide in Sub-4-Bit Quantization:\\Iterative vs.\ Single-Pass Sensitivity to Data Composition
• ICML 2026Consistency Deep Equilibrium Models
• ICLR 2026OrderDP: A Theoretically Guaranteed Lossless Dynamic Data Pruning Framework
• CVPR 2026GeoRK2 Geometry-Guided Runge-Kutta Integration for Diffusion Transformer Acceleration
• NeurIPS 2025PolarQuant: Leveraging Polar Transformation for Key Cache Quantization and Decoding Acceleration
• ICML 2025RePaViT: Scalable Vision Transformer Acceleration via Structural Reparameterization on Feedforward Network Layers
• ICLR 2025Progressive distillation induces an implicit curriculum
• CVPR 2025Adapting Dense Matching for Homography Estimation with Grid-based Acceleration
• NeurIPS 2024Provable Acceleration of Nesterov's Accelerated Gradient for Asymmetric Matrix Factorization and Linear Neural Networks
• ICML 2024Diversified Batch Selection for Training Acceleration
• ICLR 2024Embarrassingly Simple Dataset Distillation
• CVPR 2024Data-Free Quantization via Pseudo-label Filtering
• NeurIPS 2023Continuous-time Analysis of Anchor Acceleration
• ICML 2023Auxiliary Modality Learning with Generalized Curriculum Distillation
• ICLR 2023Accurate Neural Training with 4-bit Matrix Multiplications at Standard Formats
• CVPR 2023NIPQ: Noise Proxy-Based Integrated Pseudo-Quantization
• NeurIPS 2026AcceleGrad#: Adaptive Geometry-Aware Acceleration
• ICML 2026PsumQuant: In-line Post-training Partial Sum Quantizer for Energy Efficient NPU Inference
• ICLR 2026NLI : Non-uniform Linear Interpolation Approximation of Nonlinear Operations for Efficient LLMs Inference
• CVPR 2026SODA Sensitivity-Oriented Dynamic Acceleration for Diffusion Transformer

全部推理加速论文

共 53 篇，第 1 / 2 页

1. CVPR 2026Adaptive Spectral Feature Forecasting for Diffusion Sampling Acceleration
2. CVPR 2026D2Cache Second-Order Delta Caching for Higher Video Diffusion Acceleration
3. CVPR 2026Denoising as Path Planning Training-Free Acceleration of Diffusion Models with
4. CVPR 2026Fast3Dcache Training-free 3D Geometry Synthesis Acceleration
5. CVPR 2026Flow Map Distillation Without Data
6. CVPR 2026GeoRK2 Geometry-Guided Runge-Kutta Integration for Diffusion Transformer Acceleration
7. CVPR 2026Just-in-Time Training-Free Spatial Acceleration for Diffusion Transformers
8. CVPR 2026LESA Learnable Stage-Aware Predictors for Diffusion Model Acceleration
9. CVPR 2026Masked Auto-Regressive Variational Acceleration Fast Inference Makes Practical Reinforcement Learning
10. CVPR 2026ResCa Residual Caching for Diffusion Transformers Acceleration
11. CVPR 2026S2D Selective Spectral Decay for Quantization-Friendly Conditioning of Neural Activations
12. CVPR 2026Seele A Unified Acceleration Framework for Real-Time Gaussian Splatting on
13. CVPR 2026SODA Sensitivity-Oriented Dynamic Acceleration for Diffusion Transformer
14. CVPR 2026TAP A Token-Adaptive Predictor Framework for Training-Free Diffusion Acceleration
15. CVPR 2026TC-Pade Trajectory-Consistent Pade Approximation for Diffusion Acceleration
16. CVPR 2025Adapting Dense Matching for Homography Estimation with Grid-based Acceleration
17. CVPR 2025Attend to Not Attended: Structure-then-Detail Token Merging for Post-training DiT Acceleration
18. CVPR 2025Efficient ANN-Guided Distillation: Aligning Rate-based Features of Spiking Neural Networks through Hybrid Block-wise Replacement
19. CVPR 2025HOT: Hadamard-based Optimized Training
20. CVPR 2025Instant Adversarial Purification with Adversarial Consistency Distillation
21. CVPR 2025Libra-Merging: Importance-redundancy and Pruning-merging Trade-off for Acceleration Plug-in in Large Vision-Language Model
22. CVPR 2025Quantization without Tears
23. CVPR 2025RayFlow: Instance-Aware Diffusion Acceleration via Adaptive Flow Trajectories
24. CVPR 2025SharpDepth: Sharpening Metric Depth Predictions Using Diffusion Distillation
25. CVPR 20243D Paintbrush: Local Stylization of 3D Shapes with Cascaded Score Distillation
26. CVPR 2024Are Conventional SNNs Really Efficient? A Perspective from Network Quantization
27. CVPR 2024CLIP-KD: An Empirical Study of CLIP Model Distillation
28. CVPR 2024Data-Free Quantization via Pseudo-label Filtering
29. CVPR 2024Enhancing Post-training Quantization Calibration through Contrastive Learning
30. CVPR 2024Nearest is Not Dearest: Towards Practical Defense against Quantization-conditioned Backdoor Attacks
31. CVPR 2024One-step Diffusion with Distribution Matching Distillation
32. CVPR 2024PeerAiD: Improving Adversarial Distillation from a Specialized Peer Tutor
33. CVPR 2024QDFormer: Towards Robust Audiovisual Segmentation in Complex Environments with Quantization-based Semantic Decomposition
34. CVPR 2024Retraining-Free Model Quantization via One-Shot Weight-Coupling Learning
35. CVPR 2024Scaled Decoupled Distillation
36. CVPR 2024SDPose: Tokenized Pose Estimation via Circulation-Guide Self-Distillation
37. CVPR 2024Towards More Accurate Diffusion Model Acceleration with A Timestep Tuner
38. CVPR 2023Adaptive Data-Free Quantization
39. CVPR 2023Binary Latent Diffusion
40. CVPR 2023Boost Vision Transformer With GPU-Friendly Sparsity and Quantization
41. CVPR 2023DepGraph: Towards Any Structural Pruning
42. CVPR 2023DistilPose: Tokenized Pose Regression With Heatmap Distillation
43. CVPR 2023DisWOT: Student Architecture Search for Distillation WithOut Training
44. CVPR 2023Efficient RGB-T Tracking via Cross-Modality Distillation
45. CVPR 2023Generic-to-Specific Distillation of Masked Autoencoders
46. CVPR 2023Hard Sample Matters a Lot in Zero-Shot Quantization
47. CVPR 2023Network Expansion for Practical Training Acceleration
48. CVPR 2023NIPQ: Noise Proxy-Based Integrated Pseudo-Quantization
49. CVPR 2023On Distillation of Guided Diffusion Models
50. CVPR 2023PD-Quant: Post-Training Quantization Based on Prediction Difference Metric

常见问题

2023–2026 年四大会议一共有多少篇推理加速论文？

共 259 篇，按年份分别是 2023 年 54 篇、2024 年 50 篇、2025 年 64 篇、2026 年 91 篇。

NeurIPS 2026 有多少篇推理加速论文？

NeurIPS 2026 有 33 篇推理加速论文，上一届是 15 篇。

ICLR 2026 有多少篇推理加速论文？

ICLR 2026 有 17 篇推理加速论文，上一届是 18 篇。

ICML 2026 有多少篇推理加速论文？

ICML 2026 有 26 篇推理加速论文，上一届是 22 篇。

CVPR 2026 有多少篇推理加速论文？

CVPR 2026 有 15 篇推理加速论文，上一届是 9 篇。

哪个会议的推理加速论文最多？

NeurIPS，2023–2026 年共 90 篇。