Corpus ID: 231861608

Searching for Fast Model Families on Datacenter Accelerators

@article{Li2021SearchingFF,
  title={Searching for Fast Model Families on Datacenter Accelerators},
  author={Sheng Li and Mingxing Tan and R. Pang and Andrew Li and Liqun Cheng and Quoc V. Le and N. Jouppi},
  journal={ArXiv},
  year={2021},
  volume={abs/2102.05610}
}
Neural Architecture Search (NAS), together with model scaling, has shown remarkable progress in designing high accuracy and fast convolutional architecture families. However, as neither NAS nor model scaling considers sufficient hardware architecture details, they do not take full advantage of the emerging datacenter (DC) accelerators. In this paper, we search for fast and accurate CNN model families for efficient inference on DC accelerators. We first analyze DC accelerators and find that… Expand

Figures and Tables from this paper

References

SHOWING 1-10 OF 68 REFERENCES
TResNet: High Performance GPU-Dedicated Architecture
  • 21
  • PDF
Fast Hardware-Aware Neural Architecture Search
  • 7
  • PDF
SqueezeNext: Hardware-Aware Neural Network Design
  • Amir Gholami, K. Kwon, +5 authors K. Keutzer
  • Computer Science
  • 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)
  • 2018
  • 105
  • PDF
EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks
  • 1,759
  • PDF
In-datacenter performance analysis of a tensor processing unit
  • N. Jouppi, C. Young, +73 authors D. Yoon
  • Computer Science
  • 2017 ACM/IEEE 44th Annual International Symposium on Computer Architecture (ISCA)
  • 2017
  • 2,117
  • PDF
FBNet: Hardware-Aware Efficient ConvNet Design via Differentiable Neural Architecture Search
  • 422
  • PDF
ProxylessNAS: Direct Neural Architecture Search on Target Task and Hardware
  • 695
  • PDF
MnasNet: Platform-Aware Neural Architecture Search for Mobile
  • 895
  • PDF
AMC: AutoML for Model Compression and Acceleration on Mobile Devices
  • 537
  • PDF
Partial Order Pruning: For Best Speed/Accuracy Trade-Off in Neural Architecture Search
  • 47
  • PDF
...
1
2
3
4
5
...