后训练量化方法综述(特邀)

doi:10.19678/j.issn.1000-3428.0252721

摘要/Abstract

摘要：

后训练量化(PTQ)是一种高效的模型压缩方法, 它无需重新训练模型, 只需少量(或无需)无标签校准数据即可将高精度浮点模型的参数转换为低比特整数表示。该方法在显著降低存储与计算开销的同时能够最大限度地保留原始模型的推理精度, 因而受到学术界与工业界的广泛关注。从PTQ的量化步骤、方法分类、工具生态、应用进展4个维度, 系统总结PTQ的研究进展。首先, 构建了量化流程框架, 涵盖动态范围统计、量化参数计算、权重与激活量化、误差优化和模型生成等步骤; 其次, 提出一个完整的量化方法分类体系, 从量化粒度、位宽、校准方法到结构导向量化; 再次, 分析了支持PTQ规模化应用的工具生态, 探讨了其在硬件适配和工程部署中的应用价值; 最后, 总结了PTQ方法的融合与应用进展, 并指出PTQ方法在实践中面临的挑战, 尤其是跨模态一致性、极低比特语义崩塌与硬件适配等难题。这些实践挑战的总结不仅揭示了当前技术的局限性, 也为未来研究提供了重要方向。本综述为学术界与工业界提供了PTQ方法的参考框架, 助力推动人工智能在资源受限场景中的广泛应用。

关键词: 后训练量化, 后训练量化步骤, 后训练量化方法分类, 工具生态, 应用进展

Abstract:

Post-Training Quantization (PTQ) is an efficient model compression method that converts the parameters of high-precision floating-point models into low-bit integer representations without requiring retraining, using only a small amount of unlabeled calibration data. This method significantly reduces storage and computational overhead while maximizing the retention of the original model's inference accuracy; therefore, it is widely recognized and adopted in both academia and industry. This paper systematically summarizes the progress of research on PTQ from four dimensions: quantization steps, method classification, tool ecosystem, and application advancements. First, a clear framework for the quantization process is constructed, covering steps such as dynamic range statistics, quantization parameter calculation, weight and activation quantization, error optimization, and model generation. Second, a complete classification system for quantization methods is proposed, which includes quantization granularity, bit width, calibration methods, and structure-guided quantization. Third, the tool ecosystem supporting the large-scale application of PTQ is analyzed, and its value in hardware adaptation and engineering deployment is discussed. Finally, this paper summarizes the progress in the integration and application of PTQ methods and highlights practical challenges, particularly those related to cross-modal consistency, extremely low-bit semantic collapse, and hardware adaptation. These challenges not only reveal the limitations of current technologies but also provide important directions for future research. This review provides a reference framework for PTQ methods in academia and industry, thereby facilitating the widespread application of artificial intelligence in resource-constrained scenarios.

Key words: Post-Training Quantization (PTQ), PTQ steps, classification of PTQ methods, tool ecosystem, application progress

张俊娜, 王泓尊, 丁春涛. 后训练量化方法综述(特邀)[J]. 计算机工程, 2026, 52(1): 33-60.

ZHANG Junna, WANG Hongzun, DING Chuntao. Survey of Post-Training Quantization Methods (Invited)[J]. Computer Engineering, 2026, 52(1): 33-60.

https://www.ecice06.com/CN/Y2026/V52/I1/33

图/表 9

图1 PTQ方法分类导图

Fig.1 PTQ method classification diagram

图2 不同量化方法在LLaMA2-7B上的CR与PPL对比(CR基于权重量化)

Fig.2 Comparison of CR and PPL of different quantization methods on LLaMA2-7B (CR quantization based on weight)

参考文献 170

1	LE SCAO T, FAN A, AKIKI C, et al. BLOOM: a 176B-parameter open-access multilingual language model[EB/OL]. [2025-06-07]. https://arxiv.org/abs/2211.05100.
2	GALLIFANT J, FISKE A, STREKALOVA L, et al. Peer review of GPT-4 technical report and systems card. PLoS Digital Health, 2024, 3(1): e0000417. doi: 10.1371/journal.pdig.0000417
3	KARUMBUNATHAN L S. NVIDIA Jetson AGX Orin series[EB/OL]. [2025-06-07]. https://www.nvidia.cn/content/dam/en-zz/Solutions/gtcf21/jetson-orin/nvidia-jetson-agx-orin-technical-brief.pdf.
4	杨春, 张睿尧, 黄泷, 等. 深度神经网络模型量化方法综述. 工程科学学报, 2023, 45(10): 1613- 1629.
	YANG C, ZHANG R Y, HUANG L, et al. A survey of quantization methods for deep neural networks. Chinese Journal of Engineering, 2023, 45(10): 1613- 1629.
5	CHEN M Z, SHAO W Q, XU P, et al. EfficientQAT: efficient quantization-aware training for large language models[EB/OL]. [2025-06-07]. https://arxiv.org/abs/2407.11062.
6	HASAN J. Optimizing large language models through quantization: a comparative analysis of PTQ and QAT techniques[EB/OL]. [2025-06-07]. https://arxiv.org/abs/2411.06084.
7	AMINE KERKOURI M, TLIBA M, CHETOUANI A, et al. Quantization effects on neural networks perception: how would quantization change the perceptual field of vision models?[C]//Proceedings of the 13th International Conference on Image Processing Theory, Tools and Applications (IPTA). Washington D.C., USA: IEEE Press, 2024: 1-6.
8	JACOB B, KLIGYS S, CHEN B, et al. Quantization and training of neural networks for efficient integer-arithmetic-only inference[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D.C., USA: IEEE Press, 2018: 2704-2713.
9	COURBARIAUX M, HUBARA I, SOUDRY D, et al. Binarized neural networks: training deep neural networks with weights and activations constrained to +1 or -1[EB/OL]. [2025-09-01]. https://arxiv.org/abs/1602.02830.
10	咸聪慧, 王天一, 李超, 等. 基于量化的深度神经网络优化研究综述. 山东师范大学学报(自然科学版), 2024, 39(1): 21- 32.
	XIAN C H, WANG T Y, LI C, et al. Review of quantization-based deep neural network optimization research. Journal of Shandong Normal University (Natural Sciences Edition), 2024, 39(1): 21- 32.
11	CHENG Y, WANG D, ZHOU P, et al. A survey of model compression and acceleration for deep neural networks[EB/OL]. [2025-09-01]. https://arxiv.org/abs/1710.09282v1.
12	KIM S, HOOPER C, WATTANAWONG T, et al. Full stack optimization of Transformer inference: a survey[EB/OL]. [2025-09-01]. https://arxiv.org/abs/2302.14017.
13	ZHU X Y, LI J, LIU Y, et al. A survey on model compression for large language models. Transactions of the Association for Computational Linguistics, 2024, 12, 1556- 1577. doi: 10.1162/tacl_a_00704
14	QIN H T, GONG R H, LIU X L, et al. Binary neural networks: a survey. Pattern Recognition, 2020, 105, 107281. doi: 10.1016/j.patcog.2020.107281
15	ZHANG Z, GAO Y C, FAN J C, et al. SelectQ: calibration data selection for post-training quantization. Machine Intelligence Research, 2025, 22(3): 499- 510. doi: 10.1007/s11633-024-1518-0
16	HUBARA I, NAHSHAN Y, HANANI Y, et al. Improving post training neural quantization: layer-wise calibration and integer programming[EB/OL]. [2025-06-07]. https://arxiv.org/abs/2006.10518.
17	GONG R H, LIU X L, JIANG S H, et al. Differentiable soft quantization: bridging full-precision and low-bit neural networks[C]// Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). Washington D. C., USA: IEEE Press, 2019: 4852-4861.
18	LI Y H, GONG R H, TAN X, et al. BRECQ: pushing the limit of post-training quantization by block reconstruction[C]//Proceedings of the International Conference on Learning Representations. Washington D. C., USA: IEEE Press, 2021: 1-16.
19	CHICCO D, WARRENS M J, JURMAN G. The coefficient of determination R-squared is more informative than SMAPE, MAE, MAPE, MSE and RMSE in regression analysis evaluation. PeerJ Computer Science, 2021, 7, e623. doi: 10.7717/peerj-cs.623
20	郭秋丹, 濮约刚, 张启军, 等. 基于舍入误差的神经网络量化方法. 计算机工程与设计, 2024, 45(8): 2534- 2539.
	GUO Q D, PU Y G, ZHANG Q J, et al. Neural network quantization method based on round-error. Computer Engineering and Design, 2024, 45(8): 2534- 2539.
21	KYURKCHIEV N, MARKOV S. Sigmoid functions: some approximation and modelling aspects[M]. [S. l. ]: LAP Lambert Academic Publishing, 2015.
22	LIU L Y, JIANG H M, HE P C, et al. On the variance of the adaptive learning rate and beyond[C]//Proceedings of the 8th International Conference on Learning Representations. Washington D. C., USA: IEEE Press, 2020: 1-14.
23	WU D, TANG Q, ZHAO Y, et al. EasyQuant: post-training quantization via scale optimization[EB/OL]. [2025-06-07]. https://arxiv.org/abs/2006.16669.
24	DING Y F, FENG W L, CHEN C Y, et al. Reg-PTQ: regression-specialized post-training quantization for fully quantized object detector[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2024: 16174-16184.
25	WANG Z W, WU Z Y, LU J W, et al. BiDet: an efficient binarized object detector[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2020: 2049-2058.
26	CHEN P, LIU J, ZHUANG B H, et al. AQD: towards accurate quantized object detection[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2021: 104-113.
27	NIU L, LIU J W, YUAN Z H, et al. Improving posttraining quantization on object detection with task loss-guided LP metric[EB/OL]. [2025-06-07]. https://arxiv.org/abs/2304.09785.
28	SO J, LEE J, AHN D, et al. Temporal dynamic quantization for diffusion models. Advances in Neural Information Processing Systems, 2023, 36, 48686- 48698.
29	WANG C Y, WANG Z W, XU X W, et al. Towards accurate post-training quantization for diffusion models[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2024: 16026-16035.
30	LIU X C, YE M, ZHOU D Y, et al. Post-training quantization with multiple points: mixed precision without mixed precision[C]//Proceedings of the AAAI Conference on Artificial Intelligence. Palo Alto, USA: AAAI Press, 2021: 8697-8705.
31	HE Y F, LIU J, WU W J, et al. EfficientDM: efficient quantization-aware fine-tuning of low-bit diffusion models[C]//Proceedings of the 12th International Conference on Learning Representations. Washington D. C., USA: IEEE Press, 2024: 1-20.
32	HUBARA I, NAHSHAN Y, HANANI Y, et al. Accurate post training quantization with small calibration sets[C]//Proceedings of the International Conference on Machine Learning. New York, USA: ACM Press, 2021: 4466-4475.
33	WEI X Y, ZHANG Y C, ZHANG X G, et al. Outlier suppression: pushing the limit of low-bit Transformer language models. Advances in Neural Information Processing Systems, 2022, 35, 17402- 17414.
34	WEI X Y, ZHANG Y C, LI Y H, et al. Outlier suppression+: accurate quantization of large language models by equivalent and optimal shifting and scaling[C]//Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing. Stroudburg, USA: ACL Press, 2023: 1648-1665.
35	KIM N J, LEE J, KIM H. HyQ: hardware-friendly post-training quantization for CNN-Transformer hybrid networks[C]//Proceedings of the 33rd International Joint Conference on Artificial Intelligence. Jeju Island, Republic of Korea: [s.n.], 2024: 4291-4299.
36	SHANG Y Z, LIU G W, KOMPELLA R R, et al. Enhancing post-training quantization calibration through contrastive learning[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2024: 15921-15930.
37	LI X Y, LIU Y J, LIAN L, et al. Q-Diffusion: quantizing diffusion models[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). Washington D. C., USA: IEEE Press, 2023: 17489-17499.
38	SHANG Y Z, YUAN Z H, XIE B, et al. Post-training quantization on diffusion models[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2023: 1972-1981.
39	TANG S A, WANG X, CHEN H, et al. Post-training quantization with progressive calibration and activation relaxing for text-to-image diffusion models[C]//Proceedings of the European Conference on Computer Vision. Berlin, Germany: Springer, 2024: 404-420.
40	YVINEC E, DAPOGNY A, BAILLY K. Gradient-based post-training quantization: challenging the status quo[EB/OL]. [2025-06-7]. https://arxiv.org/abs/2308.07662.
41	JIANG Y F, SUN N, XIE X S, et al. ADFQ-ViT: activation-distribution-friendly post-training quantization for vision Transformers. Neural Networks, 2025, 186, 107289. doi: 10.1016/j.neunet.2025.107289
42	LI Z K, XIAO J R, YANG L W, et al. RepQ-ViT: scale reparameterization for post-training quantization of vision Transformers[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). Washington D. C., USA: IEEE Press, 2023: 17181-17190.
43	OH S, SIM H, KIM J, et al. Non-uniform step size quantization for accurate post-training quantization//Proceedings of the European Conference on Computer Vision. Berlin, Germany: Springer, 2022, 658- 673.
44	MOON J, KIM D, CHEON J, et al. Instance-aware group quantization for vision Transformers[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2024: 16132-16141.
45	RANJAN N, SAVAKIS A. LRP-QViT: mixed-precision vision Transformer quantization via layer-wise relevance propagation[EB/OL]. [2025-06-07]. https://arxiv.org/abs/2401.11243.
46	LEE J, KWON Y, PARK S, et al. Q-HyViT: post-training quantization of hybrid vision Transformers with bridge block reconstruction for IoT systems. IEEE Internet of Things Journal, 2024, 11(22): 36384- 36396. doi: 10.1109/JIOT.2024.3403844
47	YANG D W, HE N, HU X, et al. Post-training quantization for re-parameterization via coarse & fine weight splitting. Journal of Systems Architecture, 2024, 147, 103065. doi: 10.1016/j.sysarc.2024.103065
48	RYU H, LIM S, SHIM H. Memory-efficient fine-tuning for quantized diffusion model[C]//Proceedings of the European Conference on Computer Vision. Berlin, Germany: Springer, 2024: 356-372.
49	WANG H X, SHANG Y Z, YUAN Z H, et al. QuEST: low-bit diffusion model quantization via efficient selective finetuning[EB/OL]. [2025-06-07]. https://arxiv.org/abs/2402.03666.
50	LIU X W, LI Z K, XIAO J R, et al. Enhanced distribution alignment for post-training quantization of diffusion models[EB/OL]. [2025-06-07]. https://arxiv.org/abs/2401.04585.
51	YAO H Y, LI P, CAO J, et al. RAPQ: rescuing accuracy for power-of-two low-bit post-training quantization[C]//Proceedings of the 31st International Joint Conference on Artificial Intelligence. Vienna, Austria: International Joint Conferences on Artificial Intelligence Organization, 2022: 1573-1579.
52	SHOMRON G, GABBAY F, KURZUM S, et al. Post-training sparsity-aware quantization. Advances in Neural Information Processing Systems, 2021, 34, 17737- 17748.
53	WANG C B, ZHENG D D, LIU Y L, et al. Leveraging inter-layer dependency for post-training quantization. Advances in Neural Information Processing Systems, 2022, 35, 6666- 6679.
54	JEON Y, LEE C, CHO E, et al. Mr. BiQ: post-training non-uniform quantization based on minimizing the reconstruction error[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2022: 12319-12328.
55	BAI S P, CHEN J, SHEN X T, et al. Unified data-free compression: pruning and quantization without fine-tuning[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). Washington D. C., USA: IEEE Press, 2023: 5853-5862.
56	LI Y H, PANDA P. TesseraQ: ultra low-bit LLM post-training quantization with block reconstruction[EB/OL]. [2025-06-07]. https://arxiv.org/abs/2410.19103.
57	YIN J J, DONG J H, WANG Y H, et al. ModuLoRA: finetuning 2-bit LLMs on consumer GPUs by integrating with modular quantizers[EB/OL]. [2025-06-07]. https://arxiv.org/abs/2309.16119.
58	XU K, LI Z C, WANG S S, et al. PTMQ: post-training multi-bit quantization of neural networks[C]//Proceedings of the AAAI Conference on Artificial Intelligence. Palo Alto, USA: AAAI Press, 2024: 16193-16201.
59	DETTMERS T, LEWIS M, SHLEIFER S, et al. 8-bit optimizers via block-wise quantization[C]//Proceedings of the International Conference on Learning Representations. Washington D. C., USA: IEEE Press, 2022: 1-19.
60	FRANTAR E, ALISTARH D. Optimal brain compression: a framework for accurate post-training quantization and pruning. Advances in Neural Information Processing Systems, 2022, 35, 4475- 4488.
61	NAGEL M, AMJAD R A, VAN BAALEN M, et al. Up or down? Adaptive rounding for post-training quantization[C]//Proceedings of the International Conference on Machine Learning. New York, USA: ACM Press, 2020: 7197-7206.
62	WEI X Y, GONG R H, LI Y H, et al. QDROP: randomly dropping quantization for extremely low-bit post-training quantization[C]// Proceedings of the International Conference on Learning Representations. Washington D. C., USA: IEEE Press, 2022: 1-19.
63	田程, 李正杰, 陈功富, 等. 深度神经网络低比特量化方法综述. 现代信息科技, 2025, 9(10): 23-33, 38.
	TIAN C, LI Z J, CHEN G F, et al. Review on low-bit quantization methods for deep neural networks. Modern Information Technology, 2025, 9(10): 23-33, 38.
64	HE Y F, LIU L P, LIU J, et al. PTQD: accurate post-training quantization for diffusion models[C]//Proceedings of the 37th International Conference on Neural Information Processing Systems. New York, USA: ACM Press, 2023: 13237-13249.
65	BHALGAT Y, LEE J, NAGEL M, et al. LSQ: improving low-bit quantization through learnable offsets and better initialization[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW). Washington D. C., USA: IEEE Press, 2020: 2978-2985.
66	LEE J H, KIM J, KWON S J, et al. FlexRound: learnable rounding based on element-wise division for post-training quantization[C]//Proceedings of the International Conference on Machine Learning. New York, USA: ACM Press, 2023: 18913-18939.
67	KIM H B, LEE J H, YOO S, et al. MetaMix: meta-state precision searcher for mixed-precision activation quantization[C]//Proceedings of the AAAI Conference on Artificial Intelligence. Palo Alto, USA: AAAI Press, 2024: 13132-13141.
68	ZHOU S F, LI L, ZHANG X Y, et al. LiDAR-PTQ: post-training quantization for point cloud 3D object detection[C]//Proceedings of the 12th International Conference on Learning Representations. Washington D. C., USA: IEEE Press, 2024: 1-15.
69	LIN J, TANG J M, TANG H T, et al. AWQ: activation-aware weight quantization for on-device LLM compression and acceleration. Proceedings of Machine Learning and Systems, 2024, 6, 87- 100.
70	PAN J Y, WANG C C, ZHENG K F, et al. SmoothQuant+: accurate and efficient 4-bit post-training weight quantization for LLM[EB/OL]. [2025-06-08]. https://arxiv.org/abs/2312.03788.
71	WANG P Q, WANG D S, JI Y, et al. QGAN: quantized generative adversarial networks[EB/OL]. [2025-06-08]. https://arxiv.org/abs/1901.08263.
72	MA Y X, LI H X, ZHENG X W, et al. Solving oscillation problem in post-training quantization through a theoretical perspective[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2023: 7950-7959.
73	LIN Y, ZHANG T Y, SUN P Q, et al. FQ-ViT: post-training quantization for fully quantized vision Transformer[C]//Proceedings of the 31st International Joint Conference on Artificial Intelligence. Vienna, Austria: International Joint Conferences on Artificial Intelligence Organization, 2022: 1173-1179.
74	LV C T, CHEN H, GUO J Y, et al. PTQ4SAM: post-training quantization for segment anything[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2024: 15941-15951.
75	LIU Y J, YANG H R, DONG Z, et al. NoisyQuant: noisy bias-enhanced post-training activation quantization for vision Transformers[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2023: 20321-20330.
76	CHEE J, CAI Y H, KULESHOV V, et al. QuIP: 2-bit quantization of large language models with guarantees. Advances in Neural Information Processing Systems, 2023, 36, 4396- 4429.
77	ADEPU H, ZENG Z P, ZHANG L, et al. FrameQuant: flexible low-bit quantization for Transformers[C]//Proceedings of the 41st International Conference on Machine Learning. New York, USA: ACM Press, 2024: 203-227.
78	YUAN Z H, SHANG Y Z, DONG Z. PB-LLM: partially binarized large language models[C]//Proceedings of the 12th International Conference on Learning Representations. Washington D. C., USA: IEEE Press, 2024: 1-14.
79	LIN C, PENG B, LI Z Y, et al. Bit-Shrinking: limiting instantaneous sharpness for improving post-training quantization[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2023: 16196-16205.
80	WANG M Z, SUN H X, SHI J, et al. Q-YOLO: efficient inference for real-time object detection[C]//Proceedings of the Asian Conference on Pattern Recognition. Berlin, Germany: Springer, 2023: 307-321.
81	PHAM C, HOANG A D, NGUYEN C C, et al. MetaAug: meta-data augmentation for post-training quantization[C]//Proceedings of the European Conference on Computer Vision. Berlin, Germany: Springer, 2024: 236-252.
82	ZHANG X G, QIN H T, DING Y F, et al. Diversifying sample generation for accurate data-free quantization[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2021: 15658-15667.
83	GUO C, QIU Y X, LENG J W, et al. SQuant: on-the-fly data-free quantization via diagonal hessian approximation[C]//Proceedings of the International Conference on Learning Representations. Washington D. C., USA: IEEE Press, 2022: 1-18.
84	YAO Z, WU X, LI C, et al. ZeroQuant-V2: exploring post-training quantization in LLMs from comprehensive study to low rank compensation[EB/OL]. [2025-10-30]. https://arxiv.org/abs/2303.08302.
85	DETTMERS T, SVIRSCHEVSKI R A, EGIAZARIAN V, et al. SPQR: a sparse-quantized representation for near-lossless LLM weight compression[C]//Proceedings of the 12th International Conference on Learning Representations. Washington D. C., USA: IEEE Press, 2024: 1-29.
86	CAI Y H, YAO Z W, DONG Z, et al. ZeroQ: a novel zero shot quantization framework[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2020: 13169-13178.
87	FAN C X, WANG Z Q, GUO D, et al. Data-free quantization via pseudo-label filtering[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2024: 5589-5598.
88	JEON Y, LEE C, KIM H Y. GENIE: show me the data for quantization[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2023: 12064-12073.
89	MASOURIS A, SHARMA M, BOGUSZEWSKI A, et al. Post-training model quantization using GANs for synthetic data generation[EB/OL]. [2025-06-08]. https://arxiv.org/abs/2305.06052.
90	ANDREEV P, FRITZLER A. Quantization of generative adversarial networks for efficient inference: a methodological study[C]//Proceedings of the 26th International Conference on Pattern Recognition (ICPR). Washington D. C., USA: IEEE Press, 2022: 2179-2185.
91	LI Z K, MA L P, CHEN M J, et al. Patch similarity aware data-free quantization for vision Transformers[C]//Proceedings of the European Conference on Computer Vision. Berlin, Germany: Springer, 2022: 154-170.
92	RAMACHANDRAN A, KUNDU S, KRISHNA T. CLAMP-ViT: contrastive data-free learning for adaptive post-training quantization of ViTs[C]//Proceedings of the European Conference on Computer Vision. Berlin, Germany: Springer, 2024: 307-325.
93	LI J W, ZHANG T C, YEN I E H, et al. FP8-BERT: post-training quantization for Transformer[EB/OL]. [2025-06-08]. https://arxiv.org/abs/2312.05725.
94	CAO J L, CHOLAKKAL H, ANWER R M, et al. D2Det: towards high quality object detection and instance segmentation[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2020: 11482-11491.
95	杨雨迪, 葛海波, 辛世澳, 等. 融合超分辨率和特征增强的轻量化遥感图像小目标检测. 计算机工程, 2024, 50(11): 284- 296.
	YANG Y D, GE H B, XIN S A, et al. Lightweight small-object detection for remote sensing images integrating super-resolution and feature enhancement. Computer Engineering, 2024, 50(11): 284- 296.
96	张玉博, 杨帆, 郭亚, 等. 基于视觉大模型的垃圾分类轻量化算法研究. 计算机工程, 2025, 51(7): 140- 151.
	ZHANG Y B, YANG F, GUO Y, et al. Research on lightweight algorithm for garbage classification based on visual large model. Computer Engineering, 2025, 51(7): 140- 151.
97	MAKHOV D, OSTAPETS R, ZHELAVSKAYA I, et al. Towards robust full low-bit quantization of super resolution networks C]//Proceedings of the European Conference on Computer Vision. Berlin, Germany: Springer, 2024, 182- 198.
98	TU Z J, HU J, CHEN H T, et al. Toward accurate post-training quantization for image super resolution[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2023: 5856-5865.
99	TANG C, MENG Y, JIANG J C, et al. Retraining-free model quantization via one-shot weight-coupling learning[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2024: 15855-15865.
100	NAGEL M, BAALEN M V, BLANKEVOORT T, et al. Data-free quantization through weight equalization and bias correction[C]// Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). Washington D. C., USA: IEEE Press, 2019: 1325-1334.
101	DUNG H A, PHAM C, LE T, et al. Sharpness-aware data generation for zero-shot quantization[C]//Proceedings of the International Conference on Machine Learning. New York, USA: ACM Press, 2024: 12034-12045.
102	LIU J, GONG R H, WEI X Y, et al. QLLM: accurate and efficient low-bitwidth quantization for large language models[C]// Proceedings of the 12th International Conference on Learning Representations. Washington D. C., USA: IEEE Press, 2024: 1-23.
103	XIAO G X, LIN J, SEZNEC M, et al. SmoothQuant: accurate and efficient post-training quantization for large language models[C]//Proceedings of the International Conference on Machine Learning. New York, USA: ACM Press, 2023: 38087-38099.
104	YUAN Z H, NIU L, LIU J, et al. RPTQ: reorder-based post-training quantization for large language models[EB/OL]. [2025-06-08]. https://arxiv.org/abs/2304.01089.
105	LEE C H, JIN J, KIM T, et al. OWQ: outlier-aware weight quantization for efficient fine-tuning and inference of large language models[C]//Proceedings of the AAAI Conference on Artificial Intelligence. Palo Alto, USA: AAAI Press, 2024: 13355-13364.
106	KIM S, HOOPER C R C, GHOLAMI A, et al. SqueezeLLM: dense-and-sparse quantization[C]//Proceedings of the International Conference on Machine Learning. New York, USA: ACM Press, 2024: 23901-23923.
107	KIM Y J, HENRY R, FAHIM R, et al. FineQuant: unlocking efficiency with fine-grained weight-only quantization for LLMs[EB/OL]. [2025-06-08]. https://arxiv.org/abs/2308.09723.
108	YAO Z W, WU X X, LI C, et al. Exploring post-training quantization in LLMs from comprehensive study to low rank compensation[C]//Proceedings of the AAAI Conference on Artificial Intelligence. Palo Alto, USA: AAAI Press, 2024: 19377-19385.
109	DING X, LIU X Y, TU Z J, et al. CBQ: cross-block quantization for large language models[C]//Proceedings of the International Conference on Learning Representations. Washington D. C., USA: IEEE Press, 2025: 1-20.
110	YAO Z W, AMINABADI R Y, ZHANG M J, et al. ZeroQuant: efficient and affordable post-training quantization for large-scale Transformers. Advances in Neural Information Processing Systems, 2022, 35, 27168- 27183.
111	FRANTAR E, ASHKBOOS S, HOEFLER T, et al. GPTQ: accurate post-training quantization for generative pre-trained Transformers[C]//Proceedings of the 11th International Conference on Learning Representations. Washington D. C., USA: IEEE Press, 2023: 1-16.
112	ZAFRIR O, BOUDOUKH G, IZSAK P, et al. Q8BERT: quantized 8 bit BERT[C]// Proceedings of the 5th Workshop on Energy Efficient Machine Learning and Cognitive Computing-NeurIPS Edition (EMC2-NIPS). Washington D.C., USA: IEEE Press, 2019: 36-39.
113	SHEN S, DONG Z, YE J Y, et al. Q-BERT: Hessian based ultra low precision quantization of BERT[C]// Proceedings of the AAAI Conference on Artificial Intelligence. Palo Alto, USA: AAAI Press, 2020: 8815-8821.
114	LIU Z, ZHAO C, FEDOROV I, et al. SpinQuant: LLM quantization with learned rotations[C]//Proceedings of the 13th International Conference on Learning Representations. Washington D. C., USA: IEEE Press, 2024: 1-24.
115	LIN Y, TANG H, YANG S, et al. QServe: W4A8KV4 quantization and system co-design for efficient LLM serving[C]//Proceedings of the 8th Conference on Machine Learning and Systems. Washington D. C., USA: IEEE Press, 2024: 1-28.
116	LIU Z H, WANG Y H, HAN K, et al. Post-training quantization for vision Transformer. Advances in Neural Information Processing Systems, 2021, 34, 28092- 28103.
117	WU Z, CHEN J X, ZHONG H W, et al. AdaLog: post-training quantization for vision Transformers with adaptive logarithm quantizer[C]//Proceedings of the European Conference on Computer Vision. Berlin, Germany: Springer, 2024: 411-427.
118	YUAN Z H, XUE C H, CHEN Y Q, et al. PTQ4ViT: post-training quantization for vision Transformers with twin uniform quantization[C]//Proceedings of the European Conference on Computer Vision. Berlin, Germany: Springer, 2022: 191-207.
119	LIU X Y, DING X, YU L, et al. PQ-SAM: post-training quantization for segment anything model[C]//Proceedings of the European Conference on Computer Vision. Berlin, Germany: Springer, 2024: 420-437.
120	ZHONG Y S, HU J W, HUANG Y, et al. ERQ: error reduction for post-training quantization of vision Transformers[C]//Proceedings of the International Conference on Machine Learning. New York, USA: ACM Press, 2024: 61664-61680.
121	RANJAN N, SAVAKIS A. Mix-QViT: mixed-precision vision Transformer quantization driven by layer importance and quantization sensitivity[EB/OL]. [2025-06-08]. https://arxiv.org/abs/2501.06357.
122	YAO Y Z, TIAN F, CHEN J, et al. Timestep-aware correction for quantized diffusion models[C]//Proceedings of the European Conference on Computer Vision. Berlin, Germany: Springer, 2024: 215-232.
123	YANG Y W, DAI X L, WANG J L, et al. Efficient quantization strategies for latent diffusion models[EB/OL]. [2025-06-08]. https://arxiv.org/abs/2312.05431.
124	ZHAO T C, NING X F, FANG T C, et al. MixDQ: memory-efficient few-step text-to-image diffusion models with metric-decoupled mixed precision quantization[C]//Proceedings of the European Conference on Computer Vision. Berlin, Germany: Springer, 2024: 285-302.
125	PARK G, KIM M, LEE S, et al. LUT-GEMM: quantized matrix multiplication based on LUTs for efficient inference in large-scale generative language models[C]//Proceedings of the 20th International Conference on Learning Representations. Washington D. C., USA: IEEE Press, 2024: 1-18.
126	DETTMERS T, LEWIS M, BELKADA Y, et al. LLM.int8(): 8-bit matrix multiplication for Transformers at scale. Advances in Neural Information Processing Systems, 2022, 35, 30318- 30332.
127	HOOPER C, KIM S, MOHAMMADZADEH H, et al. KVQuant: towards 10 million context length LLM inference with KV cache quantization. Advances in Neural Information Processing Systems, 2024, 37, 1270- 1303.
128	YUE Y X, YUAN Z H, DUANMU H, et al. WKVQuant: quantizing weight and key/value cache for large language models gains more[EB/OL]. [2025-06-08]. https://arxiv.org/abs/2402.12065.
129	GUO C, TANG J M, HU W M, et al. OliVe: accelerating large language models via hardware-friendly outlier-victim pair quantization[C]//Proceedings of the 50th Annual International Symposium on Computer Architecture. New York, USA: ACM Press, 2023: 1-15.
130	GUO Y P, LANG Y L, REN Q Y. GPTQT: quantize large language models twice to push the efficiency[C]//Proceedings of the IEEE International Conference on Cybernetics and Intelligent Systems (CIS) and IEEE International Conference on Robotics, Automation and Mechatronics (RAM). Washington D.C., USA: IEEE Press, 2024: 368-373.
131	BAI H L, HOU L, SHANG L F, et al. Towards efficient post-training quantization of pre-trained language models. Advances in Neural Information Processing Systems, 2022, 35, 1405- 1418.
132	MA Y X, LI H X, ZHENG X W, et al. Outlier-aware slicing for post-training quantization in vision Transformer[C]//Proceedings of the 41st International Conference on Machine Learning. New York, USA: ACM Press, 2024: 33811-33825.
133	LIU J W, NIU L, YUAN Z H, et al. PD-Quant: post-training quantization based on prediction difference metric[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D.C., USA: IEEE Press, 2023: 24427-24437.
134	SHAO W, CHEN M, ZHANG Z, et al. OmniQuant: omnidirectionally calibrated quantization for large language models[C]// Proceedings of the 12th International Conference on Learning Representations. Washington D. C., USA: IEEE Press, 2024: 1-25.
135	JØRGENSEN T E. Resource-efficient language models: quantization for fast and accessible inference[EB/OL]. [2025-06-08]. https://arxiv.org/abs/2505.08620.
136	Intel. Neural compressor[EB/OL]. [2025-06-07]. https://github.com/intel/neural-compressor.
137	CHENG W H, ZHANG W W, SHEN H H, et al. Optimize weight rounding via signed gradient descent for the quantization of LLMs[C]//Proceedings of the Conference on Empirical Methods in Natural Language Processing. Stroudburg, USA: ACL Press, 2024: 11332- 11350.
138	Hugging Face. Optimum[EB/OL]. [2025-06-07]. https://github.com/huggingface/optimum.
139	SHEN Y L, SONG K T, TAN X, et al. HuggingGPT: solving AI tasks with ChatGPT and its friends in Hugging Face. Advances in Neural Information Processing Systems, 2023, 36, 38154- 38180.
140	TARAGHI M, DORCELUS G, FOUNDJEM A, et al. Deep learning model reuse in the HuggingFace community: challenges, benefit and trends[C]//Proceedings of the IEEE International Conference on Software Analysis, Evolution and Reengineering (SANER). Washington D. C., USA: IEEE Press, 2024: 512-523.
141	PaddlePaddle. PaddleSlim[EB/OL]. [2025-06-07]. https://github.com/PaddlePaddle/PaddleSlim.
142	PaddlePaddle. PaddleSlim: model Automatic Compression Tool (ACT) user guide[EB/OL]. [2025-06-07]. https://www.paddlepaddle.org.cn/documentation/docs/zh/guides/infer/paddleslim/paddle_slim_cn.html.
143	PyTorch. PyTorch native architecture optimization: torchao[EB/OL]. [2025-06-07]. https://github.com/pytorch/ao.
144	PyTorch. PyTorch quantization support documentation[EB/OL]. [2025-06-07]. https://docs.pytorch.org/docs/stable/quantization-support.html.s.
145	HUAWEI. MindSpore official english documentation[EB/OL]. [2025-06-07]. https://www.mindspore.cn/en.
146	HUAWEI. MindSpore[EB/OL]. [2025-06-07]. https://github.com/mindspore-ai/mindspore.
147	TONG Z H, DU N, SONG X B, et al. Study on MindSpore deep learning framework[C]//Proceedings of the 17th International Conference on Computational Intelligence and Security (CIS). Washington D. C., USA: IEEE Press, 2021: 183-186.
148	Google. QKeras: a quantization deep learning library for Tensorflow Keras[EB/OL]. [2025-06-07]. https://github.com/google/qkeras.
149	LORO F, PAU D, TOMASELLI V. A QKeras neural network zoo for deeply quantized imaging[C]// Proceedings of the IEEE 6th International Forum on Research and Technology for Society and Industry (RTSI). Washington D. C., USA: IEEE Press, 2021: 165-170.
150	TensorFlow. Post-training quantization guide[EB/OL]. [2025-06-07]. https://www.tensorflow.org/model_optimization/guide/quantization/post_training?hl=zh-cn.
151	PANG B, NIJKAMP E, WU Y N. Deep learning with TensorFlow: a review. Journal of Educational and Behavioral Statistics, 2020, 45(2): 227- 248. doi: 10.3102/1076998619872761
152	Alibaba. MNN: a blazing fast, lightweight deep learning framework[EB/OL]. [2025-06-07]. https://github.com/alibaba/MNN.
153	JIANG X T, WANG H, CHEN Y L, et al. MNN: a universal and efficient inference engine. Proceedings of Machine Learning and Systems, 2020, 2, 1- 13.
154	Tencent. NCNN: a high-performance neural network inference framework optimized for mobile platforms[EB/OL]. [2025-06-07]. https://github.com/Tencent/ncnn.
155	YU Y, YIN Q, ZHANG J, et al. ADMN: agent-driven modular network for dynamic parameter sharing in cooperative multi-agent reinforcement learning[C]// Proceedings of the 33rd International Joint Conference on Artificial Intelligence. Jeju Island, Republic of Korea: [s.n.], 2024: 302-310.
156	NVIDIA. NVIDIA TensorRT official documentation[EB/OL]. [2025-06-07]. https://docs.nvidia.com/deeplearning/tensorrt/.
157	NVIDIA. NVIDIA TensorRT GitHub repository[EB/OL]. [2025-06-07]. https://github.com/NVIDIA/TensorRT.
158	ZHOU Y X, GUO Z S, DONG Z, et al. TensorRT implementations of model quantization on edge SoC[C]//Proceedings of the IEEE 16th International Symposium on Embedded Multicore/Many-core Systems-on-Chip(MCSoC). Washington D. C., USA: IEEE Press, 2023: 486-493.
159	Microsoft. Quantize ONNX models[EB/OL]. [2025-06-07]. https://onnxruntime.ai/docs/performance/model-optimizations/quantization.html.
160	Microsoft. ONNX Runtime GitHub repository[EB/OL]. [2025-06-07]. https://github.com/microsoft/onnxruntime.
161	Intel. OpenVINO GitHub repository[EB/OL]. [2025-06-07]. https://github.com/openvinotoolkit/openvino.
162	Intel. OpenVINO^TM documentation[EB/OL]. [2025-06-07]. https://docs.openvino.ai/.
163	DEMIDOVSKIJ A, GORBACHEV Y, FEDOROV M, et al. OpenVINO deep learning workbench: comprehensive analysis and tuning of neural networks inference[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision Workshop (ICCVW). Washington D. C., USA: IEEE Press, 2019: 783-787.
164	LIU Z, OGUZ B, ZHAO C, et al. LLM-QAT: data-free quantization aware training for large language models[C]//Proceedings of Findings of the Association for Computational Linguistics ACL 2024. Stroudburg, USA: ACL Press, 2024: 467-484.
165	QU X Y, APONTE D, BANBURY C, et al. Automatic joint structured pruning and quantization for efficient neural network training and compression[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2025: 15234-15244.
166	GAO T X, GUO L, ZHAO S W, et al. QuantNAS: quantization-aware neural architecture search for efficient deployment on mobile device[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW). Washington D. C., USA: IEEE Press, 2024: 1704-1713.
167	XIA M, GAO T, ZENG Z, et al. Sheared LLaMA: accelerating language model pre-training via structured pruning[C]// Proceedings of the 12th International Conference on Learning Representations. Washington D. C., USA: IEEE Press, 2024: 1-25.
168	JUNG S, SON C, LEE S, et al. Learning to quantize deep networks by optimizing quantization intervals with task loss[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2019: 4345-4354.
169	ZHOU S, LI L, ZHANG X, et al. LiDAR-PTQ: post-training quantization for point cloud 3D object detection[C]//Proceedings of the 12th International Conference on Learning Representations. Washington D. C., USA: IEEE Press, 2024: 1-15.
170	XU J W, FAN J S, NAN B L, et al. ASLog: an area-efficient CNN accelerator for per-channel logarithmic post-training quantization. IEEE Transactions on Circuits and Systems I: Regular Papers, 2023, 70(12): 5380- 5393. doi: 10.1109/TCSI.2023.3315299

选择文件类型/文献管理软件名称

选择包含的内容