推理加速论文趋势 2023–2026

量化、蒸馏与 serving

统计范围：NeurIPS、ICLR、ICML、CVPR 四个会议，2023–2026 年，共 259 篇被归入该主题的论文。

主要结论

每年论文数

各会议论文数

按年份和会议统计

推理加速论文数，按年份和会议
年份NeurIPSICLRICMLCVPR合计
2023201171654
2024222131350
2025151822964
20263317261591

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

全部推理加速论文

年份全部2023202420252026
会议全部NeurIPSICLRICMLCVPR

共 90 篇，第 1 / 2 页

  1. NeurIPS 2026A Method-Class Divide in Sub-4-Bit Quantization:\\Iterative vs.\ Single-Pass Sensitivity to Data Composition
  2. NeurIPS 2026AcceleGrad#: Adaptive Geometry-Aware Acceleration
  3. NeurIPS 2026AccelEval: A Whole-Program Benchmark for LLM-Generated CPU-to-GPU Code Acceleration
  4. NeurIPS 2026Adaptive Conditional Gradient Sliding: Projection-Free and Line-Search-Free Acceleration
  5. NeurIPS 2026AlphaQ: Calibration-Free Bit Allocation for Mixture-of-Experts Quantization
  6. NeurIPS 2026Benchmark Recovery Does Not Certify a Frozen Tool-Trigger Contract After Quantization
  7. NeurIPS 2026Block Sphere Vector Quantization
  8. NeurIPS 2026Cache the Future: Training-Free Self-Revision for Diffusion Transformer Acceleration
  9. NeurIPS 2026Complexity of Classical Acceleration for $\ell_1$-Regularized PageRank
  10. NeurIPS 2026Direct Acceleration of Stochastic Root-Finding Without Variance Reduction and Regularization
  11. NeurIPS 2026Disentangling the Good From the Bad: Quantization-Induced Flips Are Not Random
  12. NeurIPS 2026Facts Don't Speak Louder than Words: The Behavioral Essence of Long-CoT Distillation
  13. NeurIPS 2026FADE: Fractional Anomalous Dynamics Extrapolation for Training-Free Diffusion Transformers Acceleration
  14. NeurIPS 2026FlashControl: One-Step Controllable Generator via Distillation-Friendly Single-Stream Teachers.
  15. NeurIPS 2026GDMD: Guiding Distribution Matching Distillation with Gradient-Based Reinforcement Learning
  16. NeurIPS 2026InQuant: In-Place Mixed-Precision KV Cache Quantization via Saliency-Aware Neighbor-Slot Reuse
  17. NeurIPS 2026Locking Pretrained Weights via Deep Low-Rank Residual Distillation
  18. NeurIPS 2026Logical Distillation of Transformer Encoders
  19. NeurIPS 2026Matched-Control Tests of Partition-Source Claims in One Routed Distillation Family
  20. NeurIPS 2026MPQ: A Message-Passing View of Post-Training Quantization
  21. NeurIPS 2026On the Memorization of Consistency Distillation for Diffusion Models
  22. NeurIPS 2026Perfect Parallelization in Mini-Batch SGD with Classical Momentum Acceleration
  23. NeurIPS 2026Provable Quantization with Randomized Hadamard Transform
  24. NeurIPS 2026RaZeR: Pushing the Limits of NVFP4 Quantization with Redundant Zero Remapping
  25. NeurIPS 2026Rethinking Gradient Approximation in Quantization: A Zeroth-Order Expectation Perspective
  26. NeurIPS 2026SHRED: Retain-Set-Free Unlearning via Self-Distillation with Logit Demotion
  27. NeurIPS 2026The Quantization Benefits of Residual-Free Transformers
  28. NeurIPS 2026Tikhonov-Stabilized Bezier Representation Forecasting for Training-free Diffusion Acceleration
  29. NeurIPS 2026Total Variation Distance Estimation in Autoregressive Models
  30. NeurIPS 2026Towards Understanding Momentum Acceleration in River-Valley Loss Landscape
  31. NeurIPS 2026Transferable SCF-Acceleration through Solver-Aligned Initialization Learning
  32. NeurIPS 2026Transporting Quantiles to Codebook: Scalable Vector Quantization without Codebook Collapse
  33. NeurIPS 2026UniPrefill: Universal Long-Context Prefill Acceleration via Block-wise Dynamic Sparsification
  34. NeurIPS 2025$\text{S}^2$Q-VDiT: Accurate Quantized Video Diffusion Transformer with Salient Data and Sparse Token Distillation
  35. NeurIPS 2025Accelerating Parallel Diffusion Model Serving with Residual Compression
  36. NeurIPS 2025Advancing Wasserstein Convergence Analysis of Score-Based Models: Insights from Discretization and Second-Order Acceleration
  37. NeurIPS 2025Breaking AR’s Sampling Bottleneck: Provable Acceleration via Diffusion Language Models
  38. NeurIPS 2025CAS-Spec: Cascade Adaptive Self-Speculative Decoding for On-the-Fly Lossless Inference Acceleration of LLMs
  39. NeurIPS 2025EAGLE-3: Scaling up Inference Acceleration of Large Language Models via Training-Time Test
  40. NeurIPS 2025EfficientVLA: Training-Free Acceleration and Compression for Vision-Language-Action Models
  41. NeurIPS 2025FALQON: Accelerating LoRA Fine-tuning with Low-Bit Floating-Point Arithmetic
  42. NeurIPS 2025FP4 All the Way: Fully Quantized Training of Large Language Models
  43. NeurIPS 2025How to build a consistency model: Learning flow maps via self-distillation
  44. NeurIPS 2025Optimistic Online-to-Batch Conversions for Accelerated Convergence and Universality
  45. NeurIPS 2025PolarQuant: Leveraging Polar Transformation for Key Cache Quantization and Decoding Acceleration
  46. NeurIPS 2025Quantization Error Propagation: Revisiting Layer-Wise Post-Training Quantization
  47. NeurIPS 2025Scaling Epidemic Inference on Contact Networks: Theory and Algorithms
  48. NeurIPS 2025Speculate Deep and Accurate: Lossless and Training-Free Acceleration for Offloaded LLMs via Substitute Speculative Decoding
  49. NeurIPS 2024Accelerated Regularized Learning in Finite N-Person Games
  50. NeurIPS 2024Acceleration Exists! Optimization Problems When Oracle Can Only Compare Objective Function Values

在 Claude、Cursor 里直接问这个趋势

接上 Paperlist MCP 后，写综述、挑方向或准备投稿时，不用离开对话框就能拿到这页的数字：每年各会议的论文数、增长倍数、涨得最快的子方向和代表论文，AI 会按你的问题整理。

可以直接复制这句话问 AI

用 paperlist 查推理加速在 NeurIPS 2023–2026 的论文趋势，哪一年涨得最快？

AI 会调用的工具：get_topic_stats {"topic": "inference-acceleration"}

两分钟接入 Paperlist MCP（Claude Code、Cursor、Codex、ChatGPT）

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。

「效率与系统」下的其他主题