收藏切换
DAQ: Divide-and-Conquer Strategy Based Adaptive Low-Bit Quantization Method for Vision Transformer
收藏切换
Qianru Lü, Jinwei Xu*, Jingfei Jiang, Dongsheng Li
Journal of Computer Research and Development | 2025, 62(6) : 1530 - 1546
Less
收藏切换
Journal of Computer Research and Development | 2025, 62(6): 1530-1546
• Chip Architecture Design •
DAQ: Divide-and-Conquer Strategy Based Adaptive Low-Bit Quantization Method for Vision Transformer
Full
Qianru Lü, Jinwei Xu*, Jingfei Jiang, Dongsheng Li
Affiliations
  • National Key Laboratory of Parallel and Distributed Computing (National University of Defense Technology), Changsha 410073
  • born in 1993. PhD candidate. Her main research interests include computer architecture and artificial intelligence

    born in 1990. PhD, assistant research fellow. Member of CCF. His main research interests include artificial intelligence and reconfigurable computing

    born in 1974. PhD, professor. Member of CCF. Her main research interests include reconfigurable computing, artificial intelligence, and computer architecture. ()

    born in 1978. PhD, professor. Member of CCF. His main research interests include parallel computing, artificial intelligence, computer architecture.()

About Author:

born in 1993. PhD candidate. Her main research interests include computer architecture and artificial intelligence

born in 1990. PhD, assistant research fellow. Member of CCF. His main research interests include artificial intelligence and reconfigurable computing

born in 1974. PhD, professor. Member of CCF. Her main research interests include reconfigurable computing, artificial intelligence, and computer architecture. ()

born in 1978. PhD, professor. Member of CCF. His main research interests include parallel computing, artificial intelligence, computer architecture.()

Published: 2025-06-02 doi: 10.7544/issn1000-1239.202550145
Outline
收藏切换

Vision Transformers (ViTs) have demonstrated remarkable success in computer vision tasks, but their complex architecture and computational demands hinder deployment on edge devices. While post-training quantization (PTQ) is widely adopted for model compression, existing PTQ methods exhibit severe performance degradation in 4-bit ultra-low-bitwidth scenarios. This work systematically addresses two fundamental limitations: 1) spatial mismatch between quantization-sensitive layers (e.g., Softmax) and compute-intensive layers (e.g., linear projections), where quantizing Softmax causes 80% accuracy loss despite contributing merely 8% computational load; 2) non-Gaussian activation distributions with hidden Gaussian-like clustering properties (97% values less than three times z-score). We propose DAQ (divide-and-conquer and adaptive quantization), a hardware-friendly PTQ method. DAQ adpots z-score-driven dynamic partitioning algorithm to separate data into normal-range and abnormal-range groups and quantizes the two groups with connected parameter. DAQ further explores hardware accelerated kernel such as tensor core to speed up quantization ViT models. Experimental results demonstrate that DAQ achieves a maximum improvement of 4.37% in ImageNet Top-1 accuracy under 4-bit quantization. In object detection tasks, its average error margin remains below 0.4% compared with the baseline and achieves a maximum improvement of 8.2%, even surpassing the full-precision model by 0.1% in specific cases, thereby realizing near-lossless low-bit-width quantization. Through hardware implementation optimization, DAQ achieves 43%~86% computational acceleration without significantly increasing computational overhead. This approach establishes a synergistic algorithm-hardware co-optimized quantization deployment paradigm for resource-constrained scenarios, effectively balancing model efficiency and precision retention.

vision Transformer (ViT)  /  post-training quantization (PTQ)  /  outlier  /  low-bit quantization  /  z-score  /  uniform correlation quantization
吕倩茹, 许金伟, 姜晶菲, 李东升. DAQ:基于分治策略的自适应Vision Transformer低位宽量化方法. 计算机研究与发展, 2025 , 62 (6) : 1530 -1546 . DOI: 10.7544/issn1000-1239.202550145
Qianru Lü, Jinwei Xu, Jingfei Jiang, Dongsheng Li. DAQ: Divide-and-Conquer Strategy Based Adaptive Low-Bit Quantization Method for Vision Transformer[J]. Journal of Computer Research and Development, 2025 , 62 (6) : 1530 -1546 . DOI: 10.7544/issn1000-1239.202550145
Year 2025 volume 62 Issue 6
115
52
Cite this Article
BibTeX
Article Info
doi: 10.7544/issn1000-1239.202550145
  • Receive Date:2025-03-01
  • Online Date:2026-11-10
  • Published:2025-06-02
Article Data
Affiliations
History
  • Received:2025-03-01
  • Revised:2025-04-10
Affiliations
    National Key Laboratory of Parallel and Distributed Computing (National University of Defense Technology), Changsha 410073
References
Share
https://castjournals.cast.org.cn/joweb/jsjyjyfz/EN/10.7544/issn1000-1239.202550145
Share to
QR

Scan QR to access full text

Cite this article
BibTeX
Citations
表12种不同金属材料的力学参数
科
Family
属数
Number of
genus
种数
Number of
species
占总种数比例
Percentage of
total species (%)
属
Genus
种数
Number of
species
占总种数比例
Percentage of total
species (%)
鹅膏菌科Amanitaceae 2 11 5.26 鹅膏菌属 Amanita 10 4.78
小菇科 Mycenaceae 2 12 5.74 丝盖伞属 Inocybe 5 2.39
多孔菌科 Polyporaceae 8 14 6.70 蜡蘑属 Laccaria 5 2.39
红菇科 Russulaceae 3 23 11.00 小皮伞属 Marasmius 6 2.87
小菇属 Mycena 11 5.26
光柄菇属 Pluteus 5 2.39
红菇属 Russula 17 8.13
栓菌属 Trametes 5 2.39
关闭全屏
  • BibTeX
  • EndNote
  • RefWorks
  • TxT