ID	squeezenet1_0
LR	0.1
Epochs	90
LR Gamma	0.1
Momentum	0.9
Batch Size	32
LR Step Size	30
Weight Decay	0.0001

ID	squeezenet1_1
LR	0.1
Epochs	90
LR Gamma	0.1
Momentum	0.9
Batch Size	32
LR Step Size	30
Weight Decay	0.0001

SqueezeNet

pytorch / vision

Last updated on Feb 12, 2021

Parameters 1 Million

FLOPs 823 Million

File Size 4.79 MB

Training Data ImageNet

Training Resources 8x NVIDIA V100 GPUs

Training Time

Training Techniques	Weight Decay, SGD with Momentum
Architecture	1x1 Convolution, Fire Module, Batch Normalization, Convolution, Dropout, Global Average Pooling, Residual Connection, ReLU, Max Pooling, Softmax
ID	squeezenet1_0
LR	0.1
Epochs	90
LR Gamma	0.1
Momentum	0.9
Batch Size	32
LR Step Size	30
Weight Decay	0.0001
SHOW MORE
SHOW LESS

Parameters 1 Million

FLOPs 352 Million

File Size 4.74 MB

Training Data ImageNet

Training Resources 8x NVIDIA V100 GPUs

Training Time

Training Techniques	Weight Decay, SGD with Momentum
Architecture	1x1 Convolution, Fire Module, Batch Normalization, Convolution, Dropout, Global Average Pooling, Residual Connection, ReLU, Max Pooling, Softmax
ID	squeezenet1_1
LR	0.1
Epochs	90
LR Gamma	0.1
Momentum	0.9
Batch Size	32
LR Step Size	30
Weight Decay	0.0001
SHOW MORE
SHOW LESS

README.md

Summary

SqueezeNet is a convolutional neural network that employs design strategies to reduce the number of parameters, notably with the use of fire modules that "squeeze" parameters using 1x1 convolutions.

How do I load this model?

To load a pretrained model:

import torchvision.models as models
squeezenet = models.squeezenet1_0(pretrained=True)

Replace the model name with the variant you want to use, e.g. squeezenet1_0. You can find the IDs in the model summaries at the top of this page.

To evaluate the model, use the image classification recipes from the library.

python train.py --test-only --model='<model_name>'

How do I train this model?

You can follow the torchvision recipe on GitHub for training a new model afresh.

Citation

@article{DBLP:journals/corr/IandolaMAHDK16,
  author    = {Forrest N. Iandola and
               Matthew W. Moskewicz and
               Khalid Ashraf and
               Song Han and
               William J. Dally and
               Kurt Keutzer},
  title     = {SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and {\textless}1MB
               model size},
  journal   = {CoRR},
  volume    = {abs/1602.07360},
  year      = {2016},
  url       = {http://arxiv.org/abs/1602.07360},
  archivePrefix = {arXiv},
  eprint    = {1602.07360},
  timestamp = {Fri, 20 Nov 2020 16:16:06 +0100},
  biburl    = {https://dblp.org/rec/journals/corr/IandolaMAHDK16.bib},
  bibsource = {dblp computer science bibliography, https://dblp.org}
}

Results

Image Classification on ImageNet

Image Classification

BENCHMARK	MODEL	METRIC NAME	METRIC VALUE	GLOBAL RANK
ImageNet	SqueezeNet 1.1	Top 1 Accuracy	58.19%	# 303
		Top 5 Accuracy	80.62%	# 303
ImageNet	SqueezeNet 1.0	Top 1 Accuracy	58.1%	# 304
		Top 5 Accuracy	80.42%	# 304