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 50 篇，第 1 / 1 页

1. CVPR 20243D Paintbrush: Local Stylization of 3D Shapes with Cascaded Score Distillation
2. CVPR 2024Are Conventional SNNs Really Efficient? A Perspective from Network Quantization
3. CVPR 2024CLIP-KD: An Empirical Study of CLIP Model Distillation
4. CVPR 2024Data-Free Quantization via Pseudo-label Filtering
5. CVPR 2024Enhancing Post-training Quantization Calibration through Contrastive Learning
6. CVPR 2024Nearest is Not Dearest: Towards Practical Defense against Quantization-conditioned Backdoor Attacks
7. CVPR 2024One-step Diffusion with Distribution Matching Distillation
8. CVPR 2024PeerAiD: Improving Adversarial Distillation from a Specialized Peer Tutor
9. CVPR 2024QDFormer: Towards Robust Audiovisual Segmentation in Complex Environments with Quantization-based Semantic Decomposition
10. CVPR 2024Retraining-Free Model Quantization via One-Shot Weight-Coupling Learning
11. CVPR 2024Scaled Decoupled Distillation
12. CVPR 2024SDPose: Tokenized Pose Estimation via Circulation-Guide Self-Distillation
13. CVPR 2024Towards More Accurate Diffusion Model Acceleration with A Timestep Tuner
14. ICLR 2024Embarrassingly Simple Dataset Distillation
15. ICLR 2024InfoBatch: Lossless Training Speed Up by Unbiased Dynamic Data Pruning
16. ICML 2024A2Q+: Improving Accumulator-Aware Weight Quantization
17. ICML 2024Accelerating Transformer Pre-training with 2:4 Sparsity
18. ICML 2024Block Acceleration Without Momentum: On Optimal Stepsizes of Block Gradient Descent for Least-Squares
19. ICML 2024Data-free Distillation of Diffusion Models with Bootstrapping
20. ICML 2024Ditto: Quantization-aware Secure Inference of Transformers upon MPC
21. ICML 2024Diversified Batch Selection for Training Acceleration
22. ICML 2024Extreme Compression of Large Language Models via Additive Quantization
23. ICML 2024KIVI: A Tuning-Free Asymmetric 2bit Quantization for KV Cache
24. ICML 2024Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Heads
25. ICML 2024Optimal Acceleration for Minimax and Fixed-Point Problems is Not Unique
26. ICML 2024Purifying Quantization-conditioned Backdoors via Layer-wise Activation Correction with Distribution Approximation
27. ICML 2024Residual Quantization with Implicit Neural Codebooks
28. ICML 2024Towards Efficient Spiking Transformer: a Token Sparsification Framework for Training and Inference Acceleration
29. NeurIPS 2024Accelerated Regularized Learning in Finite N-Person Games
30. NeurIPS 2024Acceleration Exists! Optimization Problems When Oracle Can Only Compare Objective Function Values
31. NeurIPS 2024Adaptive Depth Networks with Skippable Sub-Paths
32. NeurIPS 2024AsyncDiff: Parallelizing Diffusion Models by Asynchronous Denoising
33. NeurIPS 2024BiDM: Pushing the Limit of Quantization for Diffusion Models
34. NeurIPS 2024Constant Acceleration Flow
35. NeurIPS 2024CRONOS: Enhancing Deep Learning with Scalable GPU Accelerated Convex Neural Networks
36. NeurIPS 2024DeltaDEQ: Exploiting Heterogeneous Convergence for Accelerating Deep Equilibrium Iterations
37. NeurIPS 2024Fast Best-of-N Decoding via Speculative Rejection
38. NeurIPS 2024Faster Diffusion: Rethinking the Role of the Encoder for Diffusion Model Inference
39. NeurIPS 2024GL-NeRF: Gauss-Laguerre Quadrature Enables Training-Free NeRF Acceleration
40. NeurIPS 2024KV Cache is 1 Bit Per Channel: Efficient Large Language Model Inference with Coupled Quantization
41. NeurIPS 2024Learning-to-Cache: Accelerating Diffusion Transformer via Layer Caching
42. NeurIPS 2024Nesterov acceleration despite very noisy gradients
43. NeurIPS 2024OneBit: Towards Extremely Low-bit Large Language Models
44. NeurIPS 2024Provable Acceleration of Nesterov's Accelerated Gradient for Asymmetric Matrix Factorization and Linear Neural Networks
45. NeurIPS 2024ScaleKD: Strong Vision Transformers Could Be Excellent Teachers
46. NeurIPS 2024ShiftAddLLM: Accelerating Pretrained LLMs via Post-Training Multiplication-Less Reparameterization
47. NeurIPS 2024Speculative Decoding with CTC-based Draft Model for LLM Inference Acceleration
48. NeurIPS 2024UniFL: Improve Latent Diffusion Model via Unified Feedback Learning
49. NeurIPS 2024Universality of AdaGrad Stepsizes for Stochastic Optimization: Inexact Oracle, Acceleration and Variance Reduction
50. NeurIPS 2024VQ-Map: Bird's-Eye-View Map Layout Estimation in Tokenized Discrete Space via Vector Quantization

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。