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 64 篇，第 1 / 2 页

1. CVPR 2025Adapting Dense Matching for Homography Estimation with Grid-based Acceleration
2. CVPR 2025Attend to Not Attended: Structure-then-Detail Token Merging for Post-training DiT Acceleration
3. CVPR 2025Efficient ANN-Guided Distillation: Aligning Rate-based Features of Spiking Neural Networks through Hybrid Block-wise Replacement
4. CVPR 2025HOT: Hadamard-based Optimized Training
5. CVPR 2025Instant Adversarial Purification with Adversarial Consistency Distillation
6. CVPR 2025Libra-Merging: Importance-redundancy and Pruning-merging Trade-off for Acceleration Plug-in in Large Vision-Language Model
7. CVPR 2025Quantization without Tears
8. CVPR 2025RayFlow: Instance-Aware Diffusion Acceleration via Adaptive Flow Trajectories
9. CVPR 2025SharpDepth: Sharpening Metric Depth Predictions Using Diffusion Distillation
10. ICLR 2025Effective Interplay between Sparsity and Quantization: From Theory to Practice
11. ICLR 2025Elucidating the Preconditioning in Consistency Distillation
12. ICLR 2025Methods for Convex $(L_0,L_1)$-Smooth Optimization: Clipping, Acceleration, and Adaptivity
13. ICLR 2025Model Equality Testing: Which Model is this API Serving?
14. ICLR 2025Nesterov acceleration in benignly non-convex landscapes
15. ICLR 2025One Step Diffusion via Shortcut Models
16. ICLR 2025Optimizing $(L_0, L_1)$-Smooth Functions by Gradient Methods
17. ICLR 2025PALMBENCH: A COMPREHENSIVE BENCHMARK OF COMPRESSED LARGE LANGUAGE MODELS ON MOBILE PLATFORMS
18. ICLR 2025PEARL: Parallel Speculative Decoding with Adaptive Draft Length
19. ICLR 2025PFDiff: Training-Free Acceleration of Diffusion Models Combining Past and Future Scores
20. ICLR 2025Progressive distillation induces an implicit curriculum
21. ICLR 2025Quamba: A Post-Training Quantization Recipe for Selective State Space Models
22. ICLR 2025Restructuring Vector Quantization with the Rotation Trick
23. ICLR 2025RTop-K: Ultra-Fast Row-Wise Top-K Selection for Neural Network Acceleration on GPUs
24. ICLR 2025SageAttention: Accurate 8-Bit Attention for Plug-and-play Inference Acceleration
25. ICLR 2025Scaling FP8 training to trillion-token LLMs
26. ICLR 2025SGD with memory: fundamental properties and stochastic acceleration
27. ICLR 2025SVDQuant: Absorbing Outliers by Low-Rank Component for 4-Bit Diffusion Models
28. ICML 2025any4: Learned 4-bit Numeric Representation for LLMs
29. ICML 2025BanditSpec: Adaptive Speculative Decoding via Bandit Algorithms
30. ICML 2025COGNATE: Acceleration of Sparse Tensor Programs on Emerging Hardware using Transfer Learning
31. ICML 2025Compress then Serve: Serving Thousands of LoRA Adapters with Little Overhead
32. ICML 2025Constant Stepsize Local GD for Logistic Regression: Acceleration by Instability
33. ICML 2025Distilling the Knowledge in Data Pruning
34. ICML 2025Ensemble Distribution Distillation via Flow Matching
35. ICML 2025Fast Large Language Model Collaborative Decoding via Speculation
36. ICML 2025Learning to Keep a Promise: Scaling Language Model Decoding Parallelism with Learned Asynchronous Decoding
37. ICML 2025Matryoshka Quantization
38. ICML 2025Metadata Conditioning Accelerates Language Model Pre-training
39. ICML 2025On Exact Bit-level Reversible Transformers Without Changing Architecture
40. ICML 2025Optimizing Large Language Model Training Using FP4 Quantization
41. ICML 2025Quamba2: A Robust and Scalable Post-training Quantization Framework for Selective State Space Models
42. ICML 2025RePaViT: Scalable Vision Transformer Acceleration via Structural Reparameterization on Feedforward Network Layers
43. ICML 2025SADA: Stability-guided Adaptive Diffusion Acceleration
44. ICML 2025SageAttention2: Efficient Attention with Thorough Outlier Smoothing and Per-thread INT4 Quantization
45. ICML 2025Scalable Model Merging with Progressive Layer-wise Distillation
46. ICML 2025SKIM: Any-bit Quantization Pushing The Limits of Post-Training Quantization
47. ICML 2025Sparsing Law: Towards Large Language Models with Greater Activation Sparsity
48. ICML 2025Streamline Without Sacrifice - Squeeze out Computation Redundancy in LMM
49. ICML 2025When Dynamic Data Selection Meets Data Augmentation: Achieving Enhanced Training Acceleration
50. NeurIPS 2025$\text{S}^2$Q-VDiT: Accurate Quantized Video Diffusion Transformer with Salient Data and Sparse Token Distillation

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。