The DOI matches its Crossref record; 40 of the 40 references it cites were found in a registry or an index.
RepQ-ViT: Scale Reparameterization for Post-Training Quantization of Vision Transformers
Open accessAbstract
Post-training quantization (PTQ), which only requires a tiny dataset for calibration without end-to-end retraining, is a light and practical model compression technique. Recently, several PTQ schemes for vision transformers (ViTs) have been presented; unfortunately, they typically suffer from non-trivial accuracy degradation, especially in low-bit cases. In this paper, we propose RepQ-ViT, a novel PTQ framework for ViTs based on quantization scale reparameterization, to address the above issues. RepQ-ViT decouples the quantization and inference processes, where the former employs complex quantizers and the latter employs scale-reparameterized simplified quantizers. This ensures both accurate quantization and efficient inference, which distinguishes it from existing approaches that sacrifice quantization performance to meet the target hardware. More specifically, we focus on two components with extreme distributions: post-LayerNorm activations with severe inter-channel variation and post-Softmax activations with power-law features, and initially apply channel-wise quantization and log$\sqrt 2 $ quantization, respectively. Then, we reparameterize the scales to hardware-friendly layer-wise quantization and log2 quantization for inference, with only slight accuracy or computational costs. Extensive experiments are conducted on multiple vision tasks with different model variants, proving that RepQ-ViT, without hyperparameters and expensive reconstruction procedures, can outperform existing strong baselines and encouragingly improve the accuracy of 4-bit PTQ of ViTs to a usable level. Code is available at https://github.com/zkkli/RepQ-ViT.
Overview
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Extending this concept, RepQuant [46] advocates for a per-channel dual clipping strategy that establishes specific numerical upper and lower limits for each channel, aiming for accurate quantization while minimizing bias within the quantization space.”
From Model Quantization and Hardware Acceleration for Vision Transformers: A Comprehensive Survey (2024) · cites this paper as [46]
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“RepQ-ViT [45], RepQuant [46] and LRP-QViT [47] initially employs log √ 2 quantization. This approach is better suited for the properties of attention scores in the activations with power-law distributions. Subsequently, the scales are reparameterized to change the base to 2, enabling bit-shifting operations in inference, which are more hardware-friendly.”
From Model Quantization and Hardware Acceleration for Vision Transformers: A Comprehensive Survey (2024) · cites this paper as [45]
The cited source says what the claim attributes to it.
In the citing paper
“RepQ-ViT [45], RepQuant [46] and LRP-QViT [47] initially employs log √ 2 quantization. This approach is better suited for the properties of attention scores in the activations with power-law distributions. Subsequently, the scales are reparameterized to change the base to 2, enabling bit-shifting operations in inference, which are more hardware-friendly.”
From Model Quantization and Hardware Acceleration for Vision Transformers: A Comprehensive Survey (2024) · cites this paper as [46]
The cited source says what the claim attributes to it.
In the citing paper
“RepQ-ViT [45] proposes the Scale Reparam method for Post-LayerNorm activations involving a process that begins with applying channel-wise quantization to accurately capture their severe inter-channel variations. This initial step is followed by a critical reparameterization: transforming the channel-wise quantization scales and zero points into a more hardware-friendly layer-wise format. In”
From Model Quantization and Hardware Acceleration for Vision Transformers: A Comprehensive Survey (2024) · cites this paper as [45]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., et al. (2020) · arXiv · doi:10.48550/arxiv.2010.11929 · Cited by 47,472
Microsoft COCO: Common Objects in Context
Lin, T.-Y., Maire, M., Belongie, S., Hays, J., Perona, P., Ramanan, D., Dollár, P., & Zitnick, C. L. (2014) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-10602-1_48 · Cited by 45,460
Swin Transformer: Hierarchical Vision Transformer using Shifted Windows
Liu, Z., Lin, Y., Cao, Y., Hu, H., Wei, Y., Zhang, Z., Lin, S., & Guo, B. (2021) · 2021 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv48922.2021.00986 · Cited by 35,534
He, K., Gkioxari, G., Dollar, P., & Girshick, R. (2017) · 2017 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2017.322 · Cited by 29,137
End-to-End Object Detection with Transformers
Carion, N., Massa, F., Synnaeve, G., Usunier, N., Kirillov, A., & Zagoruyko, S. (2020) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-58452-8_13 · Cited by 18,285
Cascade R-CNN: Delving Into High Quality Object Detection
Cai, Z., & Vasconcelos, N. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00644 · Cited by 7,370
Deformable DETR: Deformable Transformers for End-to-End Object Detection
Zhu, X., Su, W., Lu, L., Li, B., Wang, X., & Dai, J. (2020) · arXiv · doi:10.48550/arxiv.2010.04159 · Cited by 6,027 · cited 2×
“In addition, ViTs have also been successfully applied to high-level applications, such as object detection [2, 38] and semantic segmentation [28].”
MethodHan (2023) · doi:10.1109/tpami.2022.3152247 · Cited by 4,682
Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference
Jacob, B., Kligys, S., Chen, B., Zhu, M., Tang, M., Howard, A., Adam, H., & Kalenichenko, D. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00286 · Cited by 4,114
Microsoft COCO: Common Objects in Context
Lin, T.-Y., Maire, M., Belongie, S., Bourdev, L., Girshick, R., Hays, J., Perona, P., Ramanan, D., Zitnick, C. L., & Dollár, P. (2014) · arXiv · doi:10.48550/arxiv.1405.0312 · Cited by 2,579
Segmenter: Transformer for Semantic Segmentation
Strudel, R., Garcia, R. A., Laptev, I., & Schmid, C. (2021) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv48922.2021.00717 · Cited by 2,369
DoReFa-Net: Training Low Bitwidth Convolutional Neural Networks with Low Bitwidth Gradients
Zhou, S., Wu, Y., Ni, Z., Zhou, X., Wen, H., & Zou, Y. (2016) · arXiv · doi:10.48550/arxiv.1606.06160 · Cited by 1,690
A Survey of Quantization Methods for Efficient Neural Network Inference
Gholami, A., Kim, S., Dong, Z., Yao, Z., Mahoney, M. W., & Keutzer, K. (2022) · Low-Power Computer Vision · doi:10.1201/9781003162810-13 · Cited by 1,219 · cited 3×
“Model quantization, which reduces model complexity by decreasing the representation precision of weights and activations, is an effective and prevalent compression approach [7, 12].”
BackgroundResultMobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformer
Mehta, S., & Rastegari, M. (2021) · arXiv · doi:10.48550/arxiv.2110.02178 · Cited by 1,110
HAQ: Hardware-Aware Automated Quantization With Mixed Precision
Wang, K., Liu, Z., Lin, Y., Lin, J., & Han, S. (2019) · 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2019.00881 · Cited by 978
Quantizing deep convolutional networks for efficient inference: A whitepaper
Krishnamoorthi, R. (2018) · arXiv · doi:10.48550/arxiv.1806.08342 · Cited by 753
PACT: Parameterized Clipping Activation for Quantized Neural Networks
Choi, J., Wang, Z., Venkataramani, S., Chuang, P. I.-J., Srinivasan, V., & Gopalakrishnan, K. (2018) · arXiv · doi:10.48550/arxiv.1805.06085 · Cited by 703
A Survey of Quantization Methods for Efficient Neural Network InferencePreprint
Gholami, A., Kim, S., Dong, Z., Yao, Z., Mahoney, M. W., & Keutzer, K. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2103.13630 · Cited by 504
Data-Free Quantization Through Weight Equalization and Bias Correction
Nagel, M., Baalen, M. V., Blankevoort, T., & Welling, M. (2019) · 2019 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2019.00141 · Cited by 500
Learned Step Size Quantization
Esser, S. K., McKinstry, J. L., Bablani, D., Appuswamy, R., & Modha, D. S. (2019) · arXiv · doi:10.48550/arxiv.1902.08153 · Cited by 415
Up or Down? Adaptive Rounding for Post-Training Quantization
Nagel, M., Amjad, R. A., van Baalen, M., Louizos, C., & Blankevoort, T. (2020) · arXiv · doi:10.48550/arxiv.2004.10568 · Cited by 252
Fully Quantized Network for Object Detection
Li, R., Wang, Y., Liang, F., Qin, H., Yan, J., & Fan, R. (2019) · 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2019.00292 · Cited by 214
Patch Slimming for Efficient Vision Transformers
Tang, Y., Han, K., Wang, Y., Xu, C., Guo, J., Xu, C., & Tao, D. (2022) · 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr52688.2022.01185 · Cited by 201
FQ-ViT: Post-Training Quantization for Fully Quantized Vision Transformer
Lin, Y., Zhang, T., Sun, P., Li, Z., & Zhou, S. (2022) · Proceedings of the Thirty-First International Joint Conference on Artificial Intelligence · doi:10.24963/ijcai.2022/164 · Cited by 198 · cited 4×
“To this end, several efforts identify the key components that limit the quantization performance of ViTs, such as LayerNorm, Softmax, and GELU, and pro-pose PTQ schemes accordingly [22, 24].”
BackgroundMethodInfluentialSelf-citationPost-Training Quantization for Vision Transformer
Liu, Z., Wang, Y., Han, K., Ma, S., & Gao, W. (2021) · arXiv · doi:10.48550/arxiv.2106.14156 · Cited by 192
BRECQ: Pushing the Limit of Post-Training Quantization by Block Reconstruction
Li, Y., Gong, R., Tan, X., Yang, Y., Hu, P., Zhang, Q., Yu, F., Wang, W., & Gu, S. (2021) · arXiv · doi:10.48550/arxiv.2102.05426 · Cited by 190
Training data-efficient image transformers & distillation through attention
Touvron, H., Cord, M., Douze, M., Massa, F., Sablayrolles, A., & Jégou, H. (2020) · arXiv · doi:10.48550/arxiv.2012.12877 · Cited by 182
PTQ4ViT: Post-Training Quantization for Vision Transformers with Twin Uniform Quantization
Yuan, Z., Xue, C., Chen, Y., Wu, Q., & Sun, G. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-19775-8_12 · Cited by 163 · cited 5×
“Although various PTQ methods for convolutional neural networks (CNNs) have been proposed in previous works with good performance, they produce disappointing results on ViTs, with more than 1% accuracy drop even in 8-bit quantization [35].”
MethodResultInfluentialLogNet: Energy-efficient neural networks using logarithmic computation
Lee, E. H., Miyashita, D., Chai, E., Murmann, B., & Wong, S. S. (2017) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp.2017.7953288 · Cited by 157
QDrop: Randomly Dropping Quantization for Extremely Low-bit Post-Training QuantizationPreprint
Wei, X., Gong, R., Li, Y., Liu, X., & Yu, F. (2022) · arXiv (Cornell University) · doi:10.48550/arxiv.2203.05740 · Cited by 157
I-ViT: Integer-only Quantization for Efficient Vision Transformer Inference
Li, Z., & Gu, Q. (2023) · 2023 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.01565 · Cited by 141 · cited 1×
“For the application granularity of the above quantizers, channel-wise quantization for weights and layer-wise quantization for activations can balance accuracy and efficiency [7, 12], and are well supported by both hardware and software [11, 34, 18], and thus have become a consensus in previous works [35, 4].”
MethodSelf-citationI-ViT: Integer-only Quantization for Efficient Vision Transformer Inference
Li, Z., & Gu, Q. (2022) · arXiv · Cited by 140
Self-citationMiniViT: Compressing Vision Transformers with Weight Multiplexing
Zhang, J., Peng, H., Wu, K., Liu, M., Xiao, B., Fu, J., & Yuan, L. (2022) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52688.2022.01183 · Cited by 135
Towards Accurate Post-Training Quantization for Vision Transformer
Ding, Y., Qin, H., Yan, Q., Chai, Z., Liu, J., Wei, X., & Liu, X. (2022) · Proceedings of the 30th ACM International Conference on Multimedia · doi:10.1145/3503161.3547826 · Cited by 79
Patch Similarity Aware Data-Free Quantization for Vision Transformers
Li, Z., Ma, L., Chen, M., Xiao, J., & Gu, Q. (2022) · Lecture Notes in Computer Science · doi:10.1007/978-3-031-20083-0_10 · Cited by 61
Self-citationPSAQ-ViT V2: Toward Accurate and General Data-Free Quantization for Vision Transformers
Li, Z., Chen, M., Xiao, J., & Gu, Q. (2024) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2023.3301007 · Cited by 41
Self-citationMulti-Dimensional Vision Transformer Compression via Dependency Guided Gaussian Process Search
Hou, Z., & Kung, S.-Y. (2022) · 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw56347.2022.00411 · Cited by 20
Multi-Dimensional Model Compression of Vision Transformer
Hou, Z., & Kung, S. (2022) · IEEE International Conference on Multimedia and Expo (ICME) · doi:10.1109/icme52920.2022.9859786 · Cited by 16
Dual-discriminator adversarial framework for data-free quantization
Li, Z., Ma, L., Long, X., Xiao, J., & Gu, Q. (2022) · Neurocomputing · doi:10.1016/j.neucom.2022.09.076 · Cited by 9
Self-citationPSAQ-ViT V2: Towards Accurate and General Data-Free Quantization for Vision Transformers
Li, Z., Chen, M., Xiao, J., & Gu, Q. (2022) · arXiv · doi:10.48550/arxiv.2209.05687 · Cited by 2
Self-citation
Research works printed but not linked: 3
- 10.5555/2999134.2999257 · DOI
- Wang, Towards accurate post-training network quantization via bitsplit and stitching, 9847
- Yao, Hawq-v3: Dyadic neural network quantization, 11875
Cited by
Sorted by citations
I-ViT: Integer-only Quantization for Efficient Vision Transformer Inference
Li, Z., & Gu, Q. (2023) · 2023 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.01565 · Cited by 141
“Despite the promising performance, ViTs’ complicated architectures with large memory footprints and computational overheads is intolerable in real-world applications [17, 43, 47, 27], especially in time/resource-constrained scenarios.”
BackgroundMethodSelf-citationNoisyQuant: Noisy Bias-Enhanced Post-Training Activation Quantization for Vision Transformers
Liu, Y., Yang, H., Dong, Z., Keutzer, K., Du, L., & Zhang, S. (2023) · 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr52729.2023.01946 · Cited by 60
ViTs as backbones: Leveraging vision transformers for feature extraction
Elharrouss, O., Himeur, Y., Mahmood, Y., Alrabaee, S., Ouamane, A., Bensaali, F., Bechqito, Y., & Chouchane, A. (2025) · Information Fusion · doi:10.1016/j.inffus.2025.102951 · Cited by 69
Saha, S., & Xu, L. (2025) · Neurocomputing · doi:10.1016/j.neucom.2025.130417 · Cited by 85
PTQ4SAM: Post-Training Quantization for Segment Anything
Lv, C., Chen, H., Guo, J., Guo, J., Guo, J., Ding, Y., Liu, X., Liu, X., & Liu, X. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.01509 · Cited by 35
Metadata
- Authors
- Li, Z., Xiao, J., Yang, L., & Gu, Q.
- Affiliations
- Chinese Academy of Sciences (CN)
- Venue
- 2023 IEEE/CVF International Conference on Computer Vision (ICCV)
- Publisher
- IEEE
- Published
- 2023-10-01
- Pages
- 17181-17190
- OpenAlex
- W4390873673
- Cited by
- 127
- References
- 38
- Citations by year
- 2026: 39 · 2025: 59 · 2024: 26 · 2023: 3
- FWCI
- 13.59
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- IEEE/CVF International Conference on Computer Vision (ICCV) (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Funders
- Research and Development
- National Natural Science Foundation of China
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Physical Sciences › Computer Science › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Data Compression Techniques
- Keywords
- post-training quantization, vision transformer, model compression, layer-wise quantization, softmax activation, low-bit quantization, inference efficiency
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
