NASNet

ID	nasnetalarge
Dropout	0.5
Crop Pct	0.911
Momentum	0.9
Image Size	331
Interpolation	bicubic
Label Smoothing	0.1
RMSProp $\epsilon$	1.0

Parameters 89 Million

FLOPs 30 Billion

File Size 339.56 MB

Training Data ImageNet

Training Resources 50x Tesla K40 GPUs

Training Time

Training Techniques	RMSProp, Weight Decay, Label Smoothing
Architecture	Depthwise Separable Convolution, Batch Normalization, ReLU, Average Pooling, Convolution, Dropout
ID	nasnetalarge
Dropout	0.5
Crop Pct	0.911
Momentum	0.9
Image Size	331
Interpolation	bicubic
Label Smoothing	0.1
RMSProp $\epsilon$	1.0
SHOW MORE
SHOW LESS

README.md

Summary

NASNet is a type of convolutional neural network discovered through neural architecture search. The building blocks consist of normal and reduction cells.

How do I load this model?

To load a pretrained model:

import timm
m = timm.create_model('nasnetalarge', pretrained=True)
m.eval()

Replace the model name with the variant you want to use, e.g. nasnetalarge. You can find the IDs in the model summaries at the top of this page.

How do I train this model?

You can follow the timm recipe scripts for training a new model afresh.

Citation

@misc{zoph2018learning,
      title={Learning Transferable Architectures for Scalable Image Recognition}, 
      author={Barret Zoph and Vijay Vasudevan and Jonathon Shlens and Quoc V. Le},
      year={2018},
      eprint={1707.07012},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}

Results

Image Classification on ImageNet

Image Classification

BENCHMARK	MODEL	METRIC NAME	METRIC VALUE	GLOBAL RANK
ImageNet	nasnetalarge	Top 1 Accuracy	82.63%	# 39
		Top 5 Accuracy	96.05%	# 39