Pith. sign in

REVIEW 5 major objections 6 minor 27 references

MicroViT: A Vision Transformer with Low Complexity Self Attention for Edge Device

T0 review · 5 major / 6 minor · reviewed 2026-08-08 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read MicroViT sets out to prove that a Vision Transformer with a single attention head and group convolutions can run on edge devices at 3.6x the speed of MobileViT-family models while keeping ImageNet accuracy competitive.

desk verdict A plausible lightweight ViT whose headline efficiency claims are undermined by unit and consistency errors in its own benchmark tables. read the letter →

arxiv 2502.05800 v1 pith:HDUAKJMN submitted 2025-02-09 cs.CV

Novendra Setyawan , Chi-Chia Sun , Mao-Hsiu Hsu , Wen-Kai Kuo , Jun-Wei Hsieh This is my paper · ORCID
classification cs.CV
keywords VisionTransformerSingle-headattentionEdgedeviceGroupconvolutionLightweightmodelImageNet-1KCOCOobjectdetectionMetaFormer
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to establish that a Vision Transformer can be made practical for phones and edge hardware without surrendering accuracy. Its central claim is that MicroViT, built around an Efficient Single Head Attention (ESHA) module, runs 3.6 times faster than the MobileViT series on a GPU and consumes roughly 40 percent less energy per image on an edge device, while keeping ImageNet-1K top-1 accuracy between 72.6 and 77.1 percent. If the measured numbers hold, the usual blocker to on-device transformers, the quadratic cost of multi-head self-attention, can be bypassed by attending over only a quarter of the channels with one head and by shrinking the spatial token grid before the dot product. The paper reports the same efficiency story on COCO object detection, where the largest MicroViT variant reaches 36.0 average precision with far fewer FLOPs than a comparable EfficientViT backbone.

What carries the argument

The engine is the ESHA block defined by Equations 1-3: a grouped 3x3 convolution splits the input into Q, K, V, and U; the value branch carries a fraction r of the channels (set to 0.215 following the single-head design in the SHViT literature); the key and value are subsampled by a depth-wise convolution with a spatial reduction ratio SR before the scaled dot-product attention; and a 1x1 convolution fuses the attended value with the activated untouched branch U. This mechanism carries the efficiency argument because it cuts both the channel dimension and the token length entering the quadratic attention, while the untouched branch preserves detail that the value branch discards. Around this block, MicroViT stacks MetaFormer residual blocks, using depth-wise convolutions as spatial mixers in the first two stages and ESHA in the last stage.

What would settle it

Re-run the Jetson Orin Nano benchmark with an external power meter, using the exact ONNX models and batch size 64, and compute average power and energy per 1000 images for MicroViT-S1 and MobileViT-XXS; if the energy-per-image ratio is not near 0.6, or if the GPU throughput gap is not close to 3.6x, the central efficiency claim fails.

Watch

Extended reading notes

Core claim

MicroViT is a three-stage pyramid MetaFormer in which the final stage replaces multi-head self-attention with ESHA. In ESHA, a single grouped 3x3 convolution produces query, key, value, and an untouched feature branch; the query and key are capped at 16 channels, the value branch uses about a quarter of the channels, and the key and value grids are spatially reduced by a depth-wise convolution before softmax attention. The authors credit this mechanism for the measured performance: MicroViT-S1 reaches 17,466 images/s on an RTX-3090 and 773 images/s on a Jetson Orin Nano with 9.1 ms latency and 19.6 J of reported energy per 1000 images, while MobileViT-XXS runs at 3,218 GPU images/s and 32.4 J per 1000 images. The paper's conclusion is that removing redundant heads and redundant channels, rather than simply shrinking the model, is what produces the 3.6x throughput gain and the energy-efficiency advantage.

Load-bearing premise

The load-bearing premise is that the throughput, latency, power, and energy numbers in the comparison tables were measured under fair, consistent conditions and that the printed values are accurate; the paper does not fully establish this because the average-power column is not physically plausible, the energy column behaves like millijoules rather than joules, and the ablation baseline's top-1 accuracy contradicts the same model's accuracy in the main ImageNet table.

Editorial extensions

If this is right

  • MicroViT-S1 achieves 72.6 percent ImageNet top-1 with 0.231 GFLOPs, beating MobileNetV2-1.0's 72.0 percent while reporting about 3.6x higher GPU throughput, so the paper claims a single attention head over a quarter of channels is enough to outperform a classic mobile CNN on both axes.
  • MicroViT-S3 matches EfficientViT-M5 at 77.1 percent top-1 while running at 9,288 images/s on the GPU versus 6,807 images/s, suggesting the ESHA design scales to higher accuracy without losing its speed edge.
  • On COCO with RetinaNet, MicroViT-S3 reaches 36.0 average precision with 159 GFLOPs, 3.3 points higher than EfficientViT-M4 while using 46.8 percent fewer FLOPs, which the paper presents as evidence that the efficiency carries over to dense prediction tasks.
  • The ablation study attributes the edge-efficiency gain mainly to group convolution: removing it raises the parameter count from 11.0M to 20.5M and drops the efficiency index from 3.07 to 1.8, so group convolution is the load-bearing component inside ESHA.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Because ESHA is described independently of the MetaFormer shell, the same attention block could be transplanted into other lightweight backbones, so the efficiency gain may transfer beyond MicroViT itself.
  • The untouched branch U, which bypasses attention entirely, suggests an explicit accuracy-versus-energy knob: sweeping the channel fraction r above 0.215 would give a testable curve of how much accuracy each additional attended channel buys.
  • An independent check of the benchmark tables should first reconcile the unit and baseline inconsistencies noted in the paper, in particular the physically implausible average-power column and the ablation baseline accuracy that differs from the main ImageNet table.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 6 minor

Summary. MicroViT proposes a lightweight vision transformer for edge deployment, centered on Efficient Single Head Attention (ESHA): a grouped convolution projects the feature map, a fraction of channels form Q, K, V, and an unaltered branch is concatenated after attention; spatial reduction (SR) lowers the QK cost. The model uses a three-stage MetaFormer structure with depthwise convolutions in early stages. On ImageNet-1K it reports S1/S2/S3 top-1 accuracies of 72.6/74.6/77.1, and on COCO with RetinaNet it reports 36.0 AP for MicroViT-S3. On a Jetson Orin Nano it reports throughput, latency, power, and energy for all models and claims 3.6x faster inference and 40% higher efficiency than the MobileViT series.

Significance. If the efficiency results survive correction, MicroViT is a plausible edge-deployable ViT: it combines known ideas (single-head attention, spatial reduction, grouped convolution) in a clean MetaFormer design, and the ImageNet-1K and COCO numbers are broadly consistent with other compact models. The architecture description (Eqs. 1-6 and Table I) is specific enough to re-implement, and the design choices are motivated. However, the paper's differentiator is the edge benchmark claim, and that claim is not currently supported by the evidence as printed: the power/energy labels are inconsistent, the 3.6x and 40% ratios cannot be reproduced from any specified baseline, and the ablation baseline disagrees with the main table. The manuscript also provides no measurement-protocol details or code, which limits independent verification. These issues are load-bearing but correctable, so a major revision is appropriate.

major comments (5)
  1. [§IV, Abstract and Table III] The headline quantitative claim, '3.6x faster inference speed and 40% higher efficiency than the MobileViT series,' is not reproducible from the reported data. On the edge-device table, the closest throughput ratio is MicroViT-S1 vs. MobileViTv2-0.5 (773/208 ≈ 3.7), and the efficiency ratio using Table II's top-1 values is MicroViT-S2 vs. MobileViTv2-0.5 ((74.6/24.3)/(70.2/31.6) ≈ 1.38, about 38%, not 40%); using the Table V baseline top-1 of 72.7 gives about 35%. The text in §IV.A also states a 3.6x GPU and 6.7x CPU speedup versus MobileNetV2-1.0, but Table II gives 17466/4527 ≈ 3.86 and 552/82 ≈ 6.73. The authors need to specify the exact baseline and re-measure or re-state the comparisons.
  2. [§IV, Table III] The power and energy columns have inconsistent units. The 'Avg Pow (W)' values are plausibly milliwatts: for MicroViT-S1, 2147 mW × 9.1 ms ≈ 19.5 mJ, which matches the listed Energy of 19.6, but the column headers say W and Joule, and 3549 W for MobileNetV2-1.0 is not physically meaningful. Consequently eta = top-1/Energy is expressed in %/mJ, not %/J, and all efficiency comparisons in Table III and the abstract need to be recomputed and re-labeled.
  3. [§IV, Table III] Throughput and latency are internally inconsistent. A throughput of 773 img/s corresponds to 1.29 ms per image, not the reported 9.1 ms latency; similar discrepancies hold for other rows. Since the abstract claims '3.6x faster inference speed,' the paper must state the batch size and measurement protocol for each metric (e.g., throughput at batch 64 versus per-image latency at batch 1) and report numbers that are mutually consistent.
  4. [§IV, Table V vs. Table II] The ablation baseline for MicroViT-S2 reports Top-1=72.7 and Param=11.0M, whereas Table II reports Top-1=74.6 and Par=10.0M for the same model. Because the Table V baseline is used to compute the efficiency eta = 3.07 and the paper's energy-efficiency claims, this discrepancy means the central efficiency result is not tied to a consistent model configuration. The authors should reconcile these values and ensure all tables refer to the same trained model.
  5. [§III.A, Table I] The channel ratio is stated inconsistently. The text says 'the optimal channel number ratio is set r = 0.215' following [17], while Table I lists ratio r = 1/4 for all MicroViT variants, and the abstract says 'a quarter of the overall channels.' Since r determines the split in Eq. (1) and the attention dimension, the paper must state the actual value used in the experiments.
minor comments (6)
  1. [§III.A, Eq. (2)] Equation (2) writes A = V · Softmax(Q^T · K); given the feature-map notation, please specify the token reshaping and the scaling factor (sqrt(d)) so the attention operation is unambiguous.
  2. [§III.A, Eq. (3)] Clarify whether sigma applies to U only or to the concatenation, and define the channel dimensions after Cat.
  3. [Table II] 'FLPs' should be 'FLOPs (G)', and the resolution column mixes 224 and 256; state whether Table III and the throughput comparisons use the same input resolution.
  4. [Table IV] The text refers to 'MicroViT-3' while the table header says 'MicroViT-S3'; use one consistent name throughout.
  5. [References] Reference [23] is attributed to 'T.-Y. Ross and G. Dollár'; this should be Lin et al., 'Focal Loss for Dense Object Detection.'
  6. [§IV.A] The sentence 'It consumes 2147 Joules' should read 'mJ' or 'mJ per image' to be consistent with Table III and with the physical scale of the measurements.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: MicroViT's efficiency and accuracy claims are empirical benchmark measurements, not reductions to fitted parameters or self-cited premises.

full rationale

The paper's derivation chain is self-contained: ESHA is defined by Equations 1-3 as group convolution plus single-head attention with spatial reduction, and the channel ratio r=0.215 is explicitly imported from external prior work (SHViT, ref [17]) as a configuration choice, not fitted to MicroViT's own results. No equation is solved from the target outputs, and no parameter is fitted to a subset of data and then renamed as a prediction. The central claims (3.6x faster inference, 40% higher efficiency) are presented as measured throughput, latency, power, and energy values in Tables II and III, and the efficiency metric eta = top-1/energy is a defined evaluation metric rather than a derived identity. The paper's self-citations (refs [2] and [7]) are contextual references to the authors' prior batik work and are never used as load-bearing justification for the architecture or its efficiency numbers. Internal inconsistencies in the benchmark tables (e.g., Avg Pow units, baseline top-1 disagreement) raise correctness and reproducibility concerns, but they are not circularity: the claims do not reduce to their inputs by construction.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The paper rests on five hand-chosen hyperparameters and four domain assumptions, plus no invented entities. The least supported assumption is the transfer of r=0.215 from SHViT to an architecture with group convolution and spatial reduction, since no ablation varies r.

free parameters (5)
  • channel ratio r = 0.215 (from SHViT [17])
    Controls the fraction of channels that go through single-head attention. It is taken as optimal from prior work, not re-derived for MicroViT with group convolution.
  • spatial reduction ratio SR = 2 for S1/S2, 1 for S3 (Table I)
    Set per variant in Table I without a reported hyperparameter search. The ablation in Table V does not clearly justify the choice.
  • group count g = 32
    Used in the 3x3 convolution in Equation 1; no ablation varies the group count.
  • QK dimension cap = 16
    Cq and Ck are set to 16 as the maximum in Section III-A; no sensitivity analysis is reported.
  • FFN expansion ratio alpha = 2
    Stated as the default expansion ratio after Equation 6; it is not tuned in the experiments.
assumptions (4)
  • domain assumption The DeiT training recipe (300 epochs, AdamW, batch 512) transfers to MicroViT and produces the reported accuracies.
    Stated in Section IV without per-model tuning or reproducibility artifacts.
  • domain assumption ONNX Runtime with BN fusion and batch size 64 is a fair benchmark configuration for every compared model.
    Only one configuration is reported; no runtime version, precision, or repeated trials are given.
  • domain assumption r=0.215 from SHViT is optimal for MicroViT's ESHA.
    Imported from [17] in Section III-A; the ablation study does not vary r.
  • domain assumption The group convolution used for local information does not discard information needed for classification or detection.
    The W/O Group ablation in Table V shows only +0.5 accuracy at nearly 1.5x FLOPs, so this premise holds only for the tested tasks and still needs external verification.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MicroViT: A Vision Transformer with Low Complexity Self Attention for Edge Device." pith.science (2026). https://pith.science/paper/HDUAKJMN

@misc{pith2026250205800,
  author       = {Pith},
  title        = {Pith review of: MicroViT: A Vision Transformer with Low Complexity Self Attention for Edge Device},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HDUAKJMN}},
  note         = {Machine review of arXiv:2502.05800}
}
read the original abstract

The Vision Transformer (ViT) has demonstrated state-of-the-art performance in various computer vision tasks, but its high computational demands make it impractical for edge devices with limited resources. This paper presents MicroViT, a lightweight Vision Transformer architecture optimized for edge devices by significantly reducing computational complexity while maintaining high accuracy. The core of MicroViT is the Efficient Single Head Attention (ESHA) mechanism, which utilizes group convolution to reduce feature redundancy and processes only a fraction of the channels, thus lowering the burden of the self-attention mechanism. MicroViT is designed using a multi-stage MetaFormer architecture, stacking multiple MicroViT encoders to enhance efficiency and performance. Comprehensive experiments on the ImageNet-1K and COCO datasets demonstrate that MicroViT achieves competitive accuracy while significantly improving 3.6 faster inference speed and reducing energy consumption with 40% higher efficiency than the MobileViT series, making it suitable for deployment in resource-constrained environments such as mobile and edge devices.

Figures

Figures reproduced from arXiv: 2502.05800 by the authors.

Figure 1
Figure 1. MicroViT conducted with 4 stage pyramid feature map MetaFormer architecture. The first two stage use Depth-Wise Convolution (DWConv) and the last two stage use Low Resolution Single Head Attention as spatial feature mixer. [10] or MobileOne [16] with Transformer architectures. They employ spatially efficient convolutions in the initial stages to minimize computational cost, while Transformer layers handle long-range… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

27 extracted references · 18 canonical work pages

  1. [17]

    Shvit: Single-head vision transformer with memory efficient macro design,

    S. Yun and Y . Ro, “Shvit: Single-head vision transformer with memory efficient macro design,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pp. 5756–5767, 2024

    work page 2024

  2. [1]

    An image is worth 16x16 words: Transformers for image recognition at scale,

    A. Dosovitskiy, “An image is worth 16x16 words: Transformers for image recognition at scale,” arXiv preprint arXiv:2010.11929 , 2020

    arXiv 2010

  3. [2]

    Multi- stage vision transformer for batik classification,

    N. Setyawan, M. N. Achmadiah, C.-C. Sun, and W.-K. Kuo, “Multi- stage vision transformer for batik classification,” in 2024 International Electronics Symposium (IES) , pp. 449–453, IEEE, 2024

    work page 2024

  4. [3]

    Swin transformer for pedestrian and occluded pedestrian detection,

    J.-A. Liang and J.-J. Ding, “Swin transformer for pedestrian and occluded pedestrian detection,” in 2024 IEEE International Symposium on Circuits and Systems (ISCAS) , pp. 1–5, IEEE, 2024

    work page 2024

  5. [4]

    Metformer: A motion enhanced transformer for multiple object tracking,

    J. Gao, K.-H. Yap, Y . Wang, K. Garg, and B. S. Han, “Metformer: A motion enhanced transformer for multiple object tracking,” in 2023 IEEE International Symposium on Circuits and Systems (ISCAS) , pp. 1– 5, IEEE, 2023

    work page 2023

  6. [5]

    Spik- ingvit: a multi-scale spiking vision transformer model for event-based object detection,

    L. Yu, H. Chen, Z. Wang, S. Zhan, J. Shao, Q. Liu, and S. Xu, “Spik- ingvit: a multi-scale spiking vision transformer model for event-based object detection,” IEEE Transactions on Cognitive and Developmental Systems, 2024

    work page 2024

  7. [6]

    Inpainting diffusion synthetic and data augment with feature keypoints for tiny partial fingerprints,

    M.-H. Hsu, Y .-C. Hsu, and C.-T. Chiu, “Inpainting diffusion synthetic and data augment with feature keypoints for tiny partial fingerprints,” IEEE Transactions on Biometrics, Behavior, and Identity Science , 2024

    work page 2024

  8. [7]

    Fpga-based batik classification using quantization aware training of mobilenet and data- flow implementation,

    N. Setyawan, C.-C. Sun, W.-K. Kuo, and M.-H. Hsu, “Fpga-based batik classification using quantization aware training of mobilenet and data- flow implementation,” in 2024 IEEE Asia Pacific Conference on Circuits and Systems (APCCAS) , pp. 306–310, IEEE, 2024

    work page 2024

Show all 27 references
  1. [8]

    Mobilevit: light-weight, general- purpose, and mobile-friendly vision transformer,

    S. Mehta and M. Rastegari, “Mobilevit: light-weight, general- purpose, and mobile-friendly vision transformer,” arXiv preprint arXiv:2110.02178, 2021

    2021 arXiv

  2. [9]

    Cvt: Introducing convolutions to vision transformers,

    H. Wu, B. Xiao, N. Codella, M. Liu, X. Dai, L. Yuan, and L. Zhang, “Cvt: Introducing convolutions to vision transformers,” in Proceedings of the IEEE/CVF international conference on computer vision , pp. 22– 31, 2021

    2021

  3. [10]

    Mobilenetv2: Inverted residuals and linear bottlenecks,

    M. Sandler, A. Howard, M. Zhu, A. Zhmoginov, and L.-C. Chen, “Mobilenetv2: Inverted residuals and linear bottlenecks,” in Proceedings of the IEEE conference on computer vision and pattern recognition , pp. 4510–4520, 2018

    2018

  4. [11]

    Pyramid vision transformer: A versatile backbone for dense prediction without convolutions,

    W. Wang, E. Xie, X. Li, D.-P. Fan, K. Song, D. Liang, T. Lu, P. Luo, and L. Shao, “Pyramid vision transformer: A versatile backbone for dense prediction without convolutions,” in Proceedings of the IEEE/CVF international conference on computer vision , pp. 568–578, 2021

    2021

  5. [12]

    Separable self-attention for mobile vision transformers,

    S. Mehta and M. Rastegari, “Separable self-attention for mobile vision transformers,” arXiv preprint arXiv:2206.02680 , 2022

    2022 arXiv

  6. [13]

    Edgenext: efficiently amalgamated cnn- transformer architecture for mobile vision applications,

    M. Maaz, A. Shaker, H. Cholakkal, S. Khan, S. W. Zamir, R. M. Anwer, and F. Shahbaz Khan, “Edgenext: efficiently amalgamated cnn- transformer architecture for mobile vision applications,” in European conference on computer vision , pp. 3–20, Springer, 2022

    2022

  7. [14]

    Fastvit: A fast hybrid vision transformer using structural reparameterization,

    P. K. A. Vasu, J. Gabriel, J. Zhu, O. Tuzel, and A. Ranjan, “Fastvit: A fast hybrid vision transformer using structural reparameterization,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 5785–5795, 2023

    2023

  8. [15]

    Rethinking vision transformers for mobilenet size and speed,

    Y . Li, J. Hu, Y . Wen, G. Evangelidis, K. Salahi, Y . Wang, S. Tulyakov, and J. Ren, “Rethinking vision transformers for mobilenet size and speed,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 16889–16900, 2023

    2023

  9. [16]

    Mobileone: An improved one millisecond mobile backbone,

    P. K. A. Vasu, J. Gabriel, J. Zhu, O. Tuzel, and A. Ranjan, “Mobileone: An improved one millisecond mobile backbone,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pp. 7907–7917, 2023

    2023

  10. [18]

    Metaformer is actually what you need for vision,

    W. Yu, M. Luo, P. Zhou, C. Si, Y . Zhou, X. Wang, J. Feng, and S. Yan, “Metaformer is actually what you need for vision,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pp. 10819–10829, 2022

    2022

  11. [19]

    Imagenet large scale visual recognition challenge,

    O. Russakovsky, J. Deng, H. Su, J. Krause, S. Satheesh, S. Ma, Z. Huang, A. Karpathy, A. Khosla, M. Bernstein, et al., “Imagenet large scale visual recognition challenge,” International journal of computer vision, vol. 115, pp. 211–252, 2015

    2015

  12. [20]

    Training data-efficient image transformers & distillation through attention,

    H. Touvron, M. Cord, M. Douze, F. Massa, A. Sablayrolles, and H. J ´egou, “Training data-efficient image transformers & distillation through attention,” in International conference on machine learning , pp. 10347–10357, PMLR, 2021

    2021

  13. [21]

    Decoupled weight decay regularization,

    I. Loshchilov, “Decoupled weight decay regularization,” arXiv preprint arXiv:1711.05101, 2017

    2017 arXiv

  14. [22]

    Microsoft coco: Common objects in context,

    T.-Y . Lin, M. Maire, S. Belongie, J. Hays, P. Perona, D. Ramanan, P. Doll ´ar, and C. L. Zitnick, “Microsoft coco: Common objects in context,” in Computer Vision–ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part V 13 , pp. 740–7...

    2014

  15. [23]

    Focal loss for dense object detection,

    T.-Y . Ross and G. Doll ´ar, “Focal loss for dense object detection,” in proceedings of the IEEE conference on computer vision and pattern recognition, pp. 2980–2988, 2017

    2017

  16. [24]

    Efficientvit: Memory efficient vision transformer with cascaded group attention,

    X. Liu, H. Peng, N. Zheng, Y . Yang, H. Hu, and Y . Yuan, “Efficientvit: Memory efficient vision transformer with cascaded group attention,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 14420–14430, 2023

    2023

  17. [25]

    MMDetection: Open mmlab detection toolbox and benchmark,

    K. Chen, J. Wang, J. Pang, Y . Cao, Y . Xiong, X. Li, S. Sun, W. Feng, Z. Liu, J. Xu, Z. Zhang, D. Cheng, C. Zhu, T. Cheng, Q. Zhao, B. Li, X. Lu, R. Zhu, Y . Wu, J. Dai, J. Wang, J. Shi, W. Ouyang, C. C. Loy, and D. Lin, “MMDetection: Open mmlab detection toolbox and benchmar...

    1906 arXiv

  18. [26]

    Run, don’t walk: chasing higher flops for faster neural networks,

    J. Chen, S.-h. Kao, H. He, W. Zhuo, S. Wen, C.-H. Lee, and S.-H. G. Chan, “Run, don’t walk: chasing higher flops for faster neural networks,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp. 12021–12031, 2023

    2023

  19. [27]

    Searching for mobilenetv3,

    A. Howard, M. Sandler, G. Chu, L.-C. Chen, B. Chen, M. Tan, W. Wang, Y . Zhu, R. Pang, V . Vasudevan, et al. , “Searching for mobilenetv3,” in Proceedings of the IEEE/CVF international conference on computer vision, pp. 1314–1324, 2019

    2019

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.