pytorchcv 0.0.41

Image classification models for PyTorch

Image classification models on PyTorch

This is a collection of image classification models. Many of them are pretrained on ImageNet-1K, CIFAR-10/100, and SVHN datasets and loaded automatically during use. All pretrained models require the same ordinary normalization. Scripts for training/evaluating/converting models are in the imgclsmob repo.

List of implemented models

• AlexNet ('One weird trick for parallelizing convolutional neural networks')
• ZFNet ('Visualizing and Understanding Convolutional Networks')
• VGG/BN-VGG ('Very Deep Convolutional Networks for Large-Scale Image Recognition')
• BN-Inception ('Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift')
• ResNet ('Deep Residual Learning for Image Recognition')
• PreResNet ('Identity Mappings in Deep Residual Networks')
• ResNeXt ('Aggregated Residual Transformations for Deep Neural Networks')
• SENet/SE-ResNet/SE-PreResNet/SE-ResNeXt ('Squeeze-and-Excitation Networks')
• IBN-ResNet/IBN-ResNeXt/IBN-DenseNet ('Two at Once: Enhancing Learning and Generalization Capacities via IBN-Net')
• AirNet/AirNeXt ('Attention Inspiring Receptive-Fields Network for Learning Invariant Representations')
• BAM-ResNet ('BAM: Bottleneck Attention Module')
• CBAM-ResNet ('CBAM: Convolutional Block Attention Module')
• ResAttNet ('Residual Attention Network for Image Classification')
• PyramidNet ('Deep Pyramidal Residual Networks')
• DiracNetV2 ('DiracNets: Training Very Deep Neural Networks Without Skip-Connections')
• ShaResNet ('ShaResNet: reducing residual network parameter number by sharing weights')
• DenseNet ('Densely Connected Convolutional Networks')
• CondenseNet ('CondenseNet: An Efficient DenseNet using Learned Group Convolutions')
• SparseNet ('Sparsely Aggregated Convolutional Networks')
• PeleeNet ('Pelee: A Real-Time Object Detection System on Mobile Devices')
• WRN ('Wide Residual Networks')
• DRN-C/DRN-D ('Dilated Residual Networks')
• DPN ('Dual Path Networks')
• DarkNet Ref/Tiny/19 ('Darknet: Open source neural networks in c')
• DarkNet-53 ('YOLOv3: An Incremental Improvement')
• ChannelNet ('ChannelNets: Compact and Efficient Convolutional Neural Networks via Channel-Wise Convolutions')
• iSQRT-COV-ResNet ('Towards Faster Training of Global Covariance Pooling Networks by Iterative Matrix Square Root Normalization')
• RevNet ('The Reversible Residual Network: Backpropagation Without Storing Activations')
• i-RevNet ('i-RevNet: Deep Invertible Networks')
• BagNet ('Approximating CNNs with Bag-of-local-Features models works surprisingly well on ImageNet')
• DLA ('Deep Layer Aggregation')
• MSDNet ('Multi-Scale Dense Networks for Resource Efficient Image Classification')
• FishNet ('FishNet: A Versatile Backbone for Image, Region, and Pixel Level Prediction')
• ESPNetv2 ('ESPNetv2: A Light-weight, Power Efficient, and General Purpose Convolutional Neural Network')
• X-DenseNet ('Deep Expander Networks: Efficient Deep Networks from Graph Theory')
• SqueezeNet/SqueezeResNet ('SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model size')
• SqueezeNext ('SqueezeNext: Hardware-Aware Neural Network Design')
• ShuffleNet ('ShuffleNet: An Extremely Efficient Convolutional Neural Network for Mobile Devices')
• ShuffleNetV2 ('ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design')
• MENet ('Merging and Evolution: Improving Convolutional Neural Networks for Mobile Applications')
• MobileNet ('MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications')
• FD-MobileNet ('FD-MobileNet: Improved MobileNet with A Fast Downsampling Strategy')
• MobileNetV2 ('MobileNetV2: Inverted Residuals and Linear Bottlenecks')
• IGCV3 ('IGCV3: Interleaved Low-Rank Group Convolutions for Efficient Deep Neural Networks')
• MnasNet ('MnasNet: Platform-Aware Neural Architecture Search for Mobile')
• DARTS ('DARTS: Differentiable Architecture Search')
• Xception ('Xception: Deep Learning with Depthwise Separable Convolutions')
• InceptionV3 ('Rethinking the Inception Architecture for Computer Vision')
• InceptionV4/InceptionResNetV2 ('Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning')
• PolyNet ('PolyNet: A Pursuit of Structural Diversity in Very Deep Networks')
• NASNet ('Learning Transferable Architectures for Scalable Image Recognition')
• PNASNet ('Progressive Neural Architecture Search')
• NIN ('Network In Network')
• RoR-3 ('Residual Networks of Residual Networks: Multilevel Residual Networks')
• RiR ('Resnet in Resnet: Generalizing Residual Architectures')
• ResDrop-ResNet ('Deep Networks with Stochastic Depth')
• Shake-Shake-ResNet ('Shake-Shake regularization')
• ShakeDrop-ResNet ('ShakeDrop Regularization for Deep Residual Learning')
• FractalNet ('FractalNet: Ultra-Deep Neural Networks without Residuals')

Installation

To use the models in your project, simply install the pytorchcv package with torch (>=0.4.1 is recommended):

pip install pytorchcv torch>=0.4.0

To enable/disable different hardware supports such as GPUs, check out PyTorch installation instructions.

Usage

Example of using a pretrained ResNet-18 model:

from pytorchcv.model_provider import get_model as ptcv_get_model
import torch
from torch.autograd import Variable

net = ptcv_get_model("resnet18", pretrained=True)
x = Variable(torch.randn(1, 3, 224, 224))
y = net(x)

Pretrained models

ImageNet-1K

Some remarks:

• Top1/Top5 are the standard 1-crop Top-1/Top-5 errors (in percents) on the validation subset of the ImageNet-1K dataset.
• FLOPs/2 is the number of FLOPs divided by two to be similar to the number of MACs.
• Remark Converted from GL model means that the model was trained on MXNet/Gluon and then converted to PyTorch.
• You may notice that quality estimations are quite different from ones for the corresponding models in other frameworks. This is due to the fact that the quality is estimated on the standard TorchVision stack of image transformations. Using OpenCV Resize transformation instead of PIL one quality evaluation results will be similar to ones for the Gluon models.

Model	Top1	Top5	Params	FLOPs/2	Remarks
AlexNet	43.48	20.93	61,100,840	714.83M	From dmlc/gluon-cv (log)
VGG-11	30.98	11.37	132,863,336	7,615.87M	From dmlc/gluon-cv (log)
VGG-13	30.07	10.75	133,047,848	11,317.65M	From dmlc/gluon-cv (log)
VGG-16	27.15	8.92	138,357,544	15,480.10M	From dmlc/gluon-cv (log)
VGG-19	26.19	8.39	143,667,240	19,642.55M	From dmlc/gluon-cv (log)
BN-VGG-11b	29.63	10.19	132,868,840	7,630.72M	From dmlc/gluon-cv (log)
BN-VGG-13b	28.41	9.63	133,053,736	11,342.14M	From dmlc/gluon-cv (log)
BN-VGG-16b	27.19	8.74	138,365,992	15,507.20M	From dmlc/gluon-cv (log)
BN-VGG-19b	26.06	8.40	143,678,248	19,672.26M	From dmlc/gluon-cv (log)
BN-Inception	25.39	8.04	11,295,240	2,048.06M	From Cadene/pretrained...pytorch (log)
ResNet-10	34.69	14.36	5,418,792	894.04M	Converted from GL model (log)
ResNet-12	33.62	13.28	5,492,776	1,126.25M	Converted from GL model (log)
ResNet-14	32.45	12.46	5,788,200	1,357.94M	Converted from GL model (log)
ResNet-16	30.49	11.18	6,968,872	1,589.34M	Converted from GL model (log)
ResNet-18 x0.25	39.62	17.85	3,937,400	270.94M	Converted from GL model (log)
ResNet-18 x0.5	33.80	13.27	5,804,296	608.70M	Converted from GL model (log)
ResNet-18 x0.75	30.40	11.06	8,476,056	1,129.45M	Converted from GL model (log)
ResNet-18	28.53	9.82	11,689,512	1,820.41M	Converted from GL model (log)
ResNet-34	25.66	8.18	21,797,672	3,672.68M	From dmlc/gluon-cv (log)
ResNet-50	22.96	6.58	25,557,032	3,877.95M	From dmlc/gluon-cv (log)
ResNet-50b	22.61	6.45	25,557,032	4,110.48M	From dmlc/gluon-cv (log)
ResNet-101	21.90	6.22	44,549,160	7,597.95M	From dmlc/gluon-cv (log)
ResNet-101b	20.88	5.61	44,549,160	7,830.48M	From dmlc/gluon-cv (log)
ResNet-152	21.01	5.50	60,192,808	11,321.85M	From dmlc/gluon-cv (log)
ResNet-152b	20.56	5.34	60,192,808	11,554.38M	From dmlc/gluon-cv (log)
PreResNet-10	35.11	14.21	5,417,128	894.19M	Converted from GL model (log)
PreResNet-12	33.86	13.48	5,491,112	1,126.40M	Converted from GL model (log)
PreResNet-14	32.64	12.39	5,786,536	1,358.09M	Converted from GL model (log)
PreResNet-16	30.53	11.08	6,967,208	1,589.49M	Converted from GL model (log)
PreResNet-18	28.43	9.72	11,687,848	1,820.56M	Converted from GL model (log)
PreResNet-34	26.23	8.41	21,796,008	3,672.83M	From dmlc/gluon-cv (log)
PreResNet-50	23.70	6.85	25,549,480	3,875.44M	From dmlc/gluon-cv (log)
PreResNet-50b	23.33	6.87	25,549,480	4,107.97M	From dmlc/gluon-cv (log)
PreResNet-101	21.74	5.91	44,541,608	7,595.44M	From dmlc/gluon-cv (log)
PreResNet-101b	21.95	6.03	44,541,608	7,827.97M	From dmlc/gluon-cv (log)
PreResNet-152	20.94	5.55	60,185,256	11,319.34M	From dmlc/gluon-cv (log)
PreResNet-152b	21.34	5.91	60,185,256	11,551.87M	From dmlc/gluon-cv (log)
PreResNet-200b	21.33	5.88	64,666,280	15,068.63M	From tornadomeet/ResNet (log)
PreResNet-269b	20.92	5.81	102,065,832	20,101.11M	From soeaver/mxnet-model (log)
ResNeXt-101 (32x4d)	21.81	6.11	44,177,704	8,003.45M	From Cadene/pretrained...pytorch (log)
ResNeXt-101 (64x4d)	21.04	5.75	83,455,272	15,500.27M	From Cadene/pretrained...pytorch (log)
SE-ResNet-50	22.47	6.40	28,088,024	3,880.49M	From Cadene/pretrained...pytorch (log)
SE-ResNet-101	21.88	5.89	49,326,872	7,602.76M	From Cadene/pretrained...pytorch (log)
SE-ResNet-152	21.48	5.76	66,821,848	11,328.52M	From Cadene/pretrained...pytorch (log)
SE-ResNeXt-50 (32x4d)	21.00	5.54	27,559,896	4,258.40M	From Cadene/pretrained...pytorch (log)
SE-ResNeXt-101 (32x4d)	19.96	5.05	48,955,416	8,008.26M	From Cadene/pretrained...pytorch (log)
SENet-154	18.62	4.61	115,088,984	20,745.78M	From Cadene/pretrained...pytorch (log)
IBN-ResNet-50	22.76	6.41	25,557,032	4,110.48M	From XingangPan/IBN-Net (log)
IBN-ResNet-101	21.29	5.61	44,549,160	7,830.48M	From XingangPan/IBN-Net (log)
IBN(b)-ResNet-50	23.64	6.86	25,558,568	4,112.89M	From XingangPan/IBN-Net (log)
IBN-ResNeXt-101 (32x4d)	20.88	5.42	44,177,704	8,003.45M	From XingangPan/IBN-Net (log)
IBN-DenseNet-121	24.47	7.25	7,978,856	2,872.13M	From XingangPan/IBN-Net (log)
IBN-DenseNet-169	23.25	6.51	14,149,480	3,403.89M	From XingangPan/IBN-Net (log)
AirNet50-1x64d (r=2)	21.84	5.90	27,425,864	4,772.11M	From soeaver/AirNet-PyTorch (log)
AirNet50-1x64d (r=16)	22.11	6.19	25,714,952	4,399.97M	From soeaver/AirNet-PyTorch (log)
AirNeXt50-32x4d (r=2)	20.87	5.51	27,604,296	5,339.58M	From soeaver/AirNet-PyTorch (log)
BAM-ResNet-50	23.14	6.58	25,915,099	4,196.09M	From Jongchan/attention-module (log)
CBAM-ResNet-50	22.38	6.05	28,089,624	4,116.97M	From Jongchan/attention-module (log)
PyramidNet-101 (a=360)	21.98	6.20	42,455,070	8,743.54M	From dyhan0920/Pyramid...PyTorch (log)
DiracNetV2-18	31.47	11.70	11,511,784	1,796.62M	From szagoruyko/diracnets (log)
DiracNetV2-34	28.75	9.93	21,616,232	3,646.93M	From szagoruyko/diracnets (log)
DenseNet-121	25.57	8.03	7,978,856	2,872.13M	From dmlc/gluon-cv (log)
DenseNet-161	22.86	6.44	28,681,000	7,793.16M	From dmlc/gluon-cv (log)
DenseNet-169	24.40	7.19	14,149,480	3,403.89M	From dmlc/gluon-cv (log)
DenseNet-201	23.10	6.63	20,013,928	4,347.15M	From dmlc/gluon-cv (log)
CondenseNet-74 (C=G=4)	26.25	8.28	4,773,944	546.06M	From ShichenLiu/CondenseNet (log)
CondenseNet-74 (C=G=8)	28.93	10.06	2,935,416	291.52M	From ShichenLiu/CondenseNet (log)
PeleeNet	31.81	11.51	2,802,248	514.87M	Converted from GL model (log)
WRN-50-2	22.53	6.41	68,849,128	11,405.42M	From szagoruyko/functional-zoo (log)
DRN-C-26	24.86	7.55	21,126,584	16,993.90M	From fyu/drn (log)
DRN-C-42	22.94	6.57	31,234,744	25,093.75M	From fyu/drn (log)
DRN-C-58	21.73	6.01	40,542,008	32,489.94M	From fyu/drn (log)
DRN-D-22	25.80	8.23	16,393,752	13,051.33M	From fyu/drn (log)
DRN-D-38	23.79	6.95	26,501,912	21,151.19M	From fyu/drn (log)
DRN-D-54	21.22	5.86	35,809,176	28,547.38M	From fyu/drn (log)
DRN-D-105	20.62	5.48	54,801,304	43,442.43M	From fyu/drn (log)
DPN-68	24.17	7.27	12,611,602	2,351.84M	From Cadene/pretrained...pytorch (log)
DPN-98	20.81	5.53	61,570,728	11,716.51M	From Cadene/pretrained...pytorch (log)
DPN-131	20.54	5.48	79,254,504	16,076.15M	From Cadene/pretrained...pytorch (log)
DarkNet Tiny	40.74	17.84	1,042,104	500.85M	Converted from GL model (log)
DarkNet Ref	38.58	17.18	7,319,416	367.59M	Converted from GL model (log)
DarkNet-53	21.75	5.64	41,609,928	7,133.86M	From dmlc/gluon-cv (log)
i-RevNet-301	25.98	8.41	125,120,356	14,453.87M	From jhjacobsen/pytorch-i-revnet (log)
BagNet-9	53.61	29.61	15,688,744	16,049.19M	From wielandbrendel/bag...models (log)
BagNet-17	41.20	18.84	16,213,032	15,768.77M	From wielandbrendel/bag...models (log)
BagNet-33	33.34	13.01	18,310,184	16,371.52M	From wielandbrendel/bag...models (log)
DLA-34	25.36	7.94	15,742,104	3,071.37M	From ucbdrive/dla (log)
DLA-46-C	35.12	13.71	1,301,400	585.45M	From ucbdrive/dla (log)
DLA-X-46-C	34.02	13.02	1,068,440	546.72M	From ucbdrive/dla (log)
DLA-60	22.98	6.69	22,036,632	4,255.49M	From ucbdrive/dla (log)
DLA-X-60	21.76	5.98	17,352,344	3,543.68M	From ucbdrive/dla (log)
DLA-X-60-C	32.09	11.57	1,319,832	596.06M	From ucbdrive/dla (log)
DLA-102	21.97	6.05	33,268,888	7,190.95M	From ucbdrive/dla (log)
DLA-X-102	21.49	5.77	26,309,272	5,884.94M	From ucbdrive/dla (log)
DLA-X2-102	20.55	5.36	41,282,200	9,340.61M	From ucbdrive/dla (log)
DLA-169	21.29	5.66	53,389,720	11,593.20M	From ucbdrive/dla (log)
FishNet-150	21.97	6.04	24,959,400	6,435.05M	From kevin-ssy/FishNet (log)
ESPNetv2 x0.5	42.32	20.15	1,241,332	35.36M	From sacmehta/ESPNetv2 (log)
ESPNetv2 x1.0	33.92	13.45	1,670,072	98.09M	From sacmehta/ESPNetv2 (log)
ESPNetv2 x1.25	32.06	12.18	1,965,440	138.18M	From sacmehta/ESPNetv2 (log)
ESPNetv2 x1.5	30.83	11.29	2,314,856	185.77M	From sacmehta/ESPNetv2 (log)
ESPNetv2 x2.0	27.94	9.61	3,498,136	306.93M	From sacmehta/ESPNetv2 (log)
SqueezeNet v1.0	39.29	17.66	1,248,424	823.67M	Converted from GL model (log)
SqueezeNet v1.1	39.31	17.72	1,235,496	352.02M	Converted from GL model (log)
SqueezeResNet v1.0	39.77	18.09	1,248,424	823.67M	Converted from GL model (log)
SqueezeResNet v1.1	40.09	18.21	1,235,496	352.02M	Converted from GL model (log)
1.0-SqNxt-23	42.51	19.06	724,056	287.28M	Converted from GL model (log)
1.0-SqNxt-23v5	40.77	17.85	921,816	285.82M	Converted from GL model (log)
1.5-SqNxt-23	34.89	13.50	1,511,824	552.39M	Converted from GL model (log)
1.5-SqNxt-23v5	33.81	13.01	1,953,616	550.97M	Converted from GL model (log)
2.0-SqNxt-23	30.62	11.00	2,583,752	898.48M	Converted from GL model (log)
2.0-SqNxt-23v5	29.63	10.66	3,366,344	897.60M	Converted from GL model (log)
ShuffleNet x0.25 (g=1)	62.44	37.29	209,746	12.35M	Converted from GL model (log)
ShuffleNet x0.25 (g=3)	61.74	36.53	305,902	13.09M	Converted from GL model (log)
ShuffleNet x0.5 (g=1)	46.59	22.61	534,484	41.16M	Converted from GL model (log)
ShuffleNet x0.5 (g=3)	44.16	20.80	718,324	41.70M	Converted from GL model (log)
ShuffleNet x0.75 (g=1)	39.58	17.11	975,214	86.42M	Converted from GL model (log)
ShuffleNet x0.75 (g=3)	38.20	16.50	1,238,266	85.82M	Converted from GL model (log)
ShuffleNet x1.0 (g=1)	34.93	13.89	1,531,936	148.13M	Converted from GL model (log)
ShuffleNet x1.0 (g=2)	34.25	13.63	1,733,848	147.60M	Converted from GL model (log)
ShuffleNet x1.0 (g=3)	34.39	13.48	1,865,728	145.46M	Converted from GL model (log)
ShuffleNet x1.0 (g=4)	34.19	13.35	1,968,344	143.33M	Converted from GL model (log)
ShuffleNet x1.0 (g=8)	34.06	13.42	2,434,768	150.76M	Converted from GL model (log)
ShuffleNetV2 x0.5	40.99	18.65	1,366,792	43.31M	Converted from GL model (log)
ShuffleNetV2 x1.0	31.44	11.63	2,278,604	149.72M	Converted from GL model (log)
ShuffleNetV2 x1.5	32.82	12.69	4,406,098	320.77M	Converted from GL model (log)
ShuffleNetV2 x2.0	32.45	12.49	7,601,686	595.84M	Converted from GL model (log)
ShuffleNetV2b x0.5	40.29	18.22	1,366,792	43.31M	Converted from GL model (log)
ShuffleNetV2b x1.0	30.62	11.25	2,279,760	150.62M	Converted from GL model (log)
ShuffleNetV2b x1.5	27.31	9.11	4,410,194	323.98M	Converted from GL model (log)
ShuffleNetV2b x2.0	25.58	8.34	7,611,290	603.37M	Converted from GL model (log)
108-MENet-8x1 (g=3)	43.94	20.76	654,516	42.68M	Converted from GL model (log)
128-MENet-8x1 (g=4)	42.43	19.59	750,796	45.98M	Converted from GL model (log)
160-MENet-8x1 (g=8)	43.84	20.84	850,120	45.63M	Converted from GL model (log)
228-MENet-12x1 (g=3)	34.11	13.16	1,806,568	152.93M	Converted from GL model (log)
256-MENet-12x1 (g=4)	32.65	12.52	1,888,240	150.65M	Converted from GL model (log)
348-MENet-12x1 (g=3)	28.24	9.58	3,368,128	312.00M	Converted from GL model (log)
352-MENet-12x1 (g=8)	31.56	12.00	2,272,872	157.35M	Converted from GL model (log)
456-MENet-24x1 (g=3)	25.32	7.99	5,304,784	567.90M	Converted from GL model (log)
MobileNet x0.25	46.26	22.49	470,072	44.09M	Converted from GL model (log)
MobileNet x0.5	34.15	13.55	1,331,592	155.42M	Converted from GL model (log)
MobileNet x0.75	30.14	10.76	2,585,560	333.99M	Converted from GL model (log)
MobileNet x1.0	26.61	8.95	4,231,976	579.80M	Converted from GL model (log)
FD-MobileNet x0.25	55.86	30.98	383,160	12.95M	Converted from GL model (log)
FD-MobileNet x0.5	43.13	20.15	993,928	41.84M	Converted from GL model (log)
FD-MobileNet x0.75	38.42	16.41	1,833,304	86.68M	Converted from GL model (log)
FD-MobileNet x1.0	34.23	13.38	2,901,288	147.46M	Converted from GL model (log)
MobileNetV2 x0.25	48.34	24.51	1,516,392	34.24M	Converted from GL model (log)
MobileNetV2 x0.5	35.98	14.93	1,964,736	100.13M	Converted from GL model (log)
MobileNetV2 x0.75	30.17	10.82	2,627,592	198.50M	Converted from GL model (log)
MobileNetV2 x1.0	26.97	8.87	3,504,960	329.36M	Converted from GL model (log)
IGCV3 x0.25	53.70	28.71	1,534,020	41.29M	Converted from GL model (log)
IGCV3 x0.5	39.75	17.32	1,985,528	111.12M	Converted from GL model (log)
IGCV3 x0.75	31.05	11.40	2,638,084	210.95M	Converted from GL model (log)
IGCV3 x1.0	27.91	9.20	3,491,688	340.79M	Converted from GL model (log)
MnasNet	31.58	11.74	4,308,816	317.67M	From zeusees/Mnasnet...Model (log)
DARTS	26.70	8.74	4,718,752	539.86M	From quark0/darts (log)
Xception	20.97	5.49	22,855,952	8,403.63M	From Cadene/pretrained...pytorch (log)
InceptionV3	21.12	5.65	23,834,568	5,743.06M	From dmlc/gluon-cv (log)
InceptionV4	20.64	5.29	42,679,816	12,304.93M	From Cadene/pretrained...pytorch (log)
InceptionResNetV2	19.93	4.90	55,843,464	13,188.64M	From Cadene/pretrained...pytorch (log)
PolyNet	19.10	4.52	95,366,600	34,821.34M	From Cadene/pretrained...pytorch (log)
NASNet-A 4@1056	25.68	8.16	5,289,978	584.90M	From Cadene/pretrained...pytorch (log)
NASNet-A 6@4032	18.14	4.21	88,753,150	23,976.44M	From Cadene/pretrained...pytorch (log)
PNASNet-5-Large	17.88	4.28	86,057,668	25,140.77M	From Cadene/pretrained...pytorch (log)

CIFAR-10

Model	Error, %	Params	FLOPs/2	Remarks
NIN	7.43	966,986	222.97M	Converted from GL model (log)
ResNet-20	5.97	272,474	41.29M	Converted from GL model (log)
ResNet-56	4.52	855,770	127.06M	Converted from GL model (log)
ResNet-110	3.69	1,730,714	255.70M	Converted from GL model (log)
ResNet-164(BN)	3.68	1,704,154	255.31M	Converted from GL model (log)
ResNet-1001	3.28	10,328,602	1,536.40M	Converted from GL model (log)
ResNet-1202	3.53	19,424,026	2,857.17M	Converted from GL model (log)
PreResNet-20	6.51	272,282	41.27M	Converted from GL model (log)
PreResNet-56	4.49	855,578	127.03M	Converted from GL model (log)
PreResNet-110	3.86	1,730,522	255.68M	Converted from GL model (log)
PreResNet-164(BN)	3.64	1,703,258	255.08M	Converted from GL model (log)
PreResNet-1001	2.65	10,327,706	1,536.18M	Converted from GL model (log)
PreResNet-1202	3.39	19,423,834	2,857.14M	Converted from GL model (log)
ResNeXt-29 (32x4d)	3.15	4,775,754	780.55M	Converted from GL model (log)
ResNeXt-29 (16x64d)	2.41	68,155,210	10,709.34M	Converted from GL model (log)
PyramidNet-110 (a=48)	3.72	1,772,706	408.37M	Converted from GL model (log)
PyramidNet-110 (a=84)	2.98	3,904,446	778.15M	Converted from GL model (log)
PyramidNet-110 (a=270)	2.51	28,485,477	4,730.60M	Converted from GL model (log)
PyramidNet-164 (a=270, BN)	2.42	27,216,021	4,608.81M	Converted from GL model (log)
DenseNet-40 (k=12)	5.61	599,050	210.80M	Converted from GL model (log)
DenseNet-BC-40 (k=12)	6.43	176,122	74.89M	Converted from GL model (log)
DenseNet-BC-40 (k=24)	4.52	690,346	293.09M	Converted from GL model (log)
DenseNet-BC-40 (k=36)	4.04	1,542,682	654.60M	Converted from GL model (log)
DenseNet-100 (k=12)	3.66	4,068,490	1,353.55M	Converted from GL model (log)
DenseNet-100 (k=24)	3.13	16,114,138	5,354.19M	Converted from GL model (log)
DenseNet-BC-100 (k=12)	4.16	769,162	298.45M	Converted from GL model (log)
X-DenseNet-BC-40-2 (k=24)	5.31	690,346	293.09M	Converted from GL model (log)
X-DenseNet-BC-40-2 (k=36)	4.37	1,542,682	654.60M	Converted from GL model (log)
WRN-16-10	2.93	17,116,634	2,414.04M	Converted from GL model (log)
WRN-28-10	2.39	36,479,194	5,246.98M	Converted from GL model (log)
WRN-40-8	2.37	35,748,314	5,176.90M	Converted from GL model (log)
RoR-3-56	5.43	762,746	113.43M	Converted from GL model (log)
RoR-3-110	4.35	1,637,690	242.07M	Converted from GL model (log)
Shake-Shake-ResNet-20-2x16d	5.15	541,082	81.78M	Converted from GL model (log)
Shake-Shake-ResNet-26-2x32d	3.17	2,923,162	428.89M	Converted from GL model (log)

CIFAR-100

Model	Error, %	Params	FLOPs/2	Remarks
NIN	28.39	984,356	224.08M	Converted from GL model (log)
ResNet-20	29.64	278,324	41.30M	Converted from GL model (log)
ResNet-56	24.88	861,620	127.06M	Converted from GL model (log)
ResNet-110	22.80	1,736,564	255.71M	Converted from GL model (log)
ResNet-164(BN)	20.44	1,727,284	255.33M	Converted from GL model (log)
ResNet-1001	19.79	10,351,732	1,536.43M	Converted from GL model (log)
PreResNet-20	30.22	278,132	41.28M	Converted from GL model (log)
PreResNet-56	25.05	861,428	127.04M	Converted from GL model (log)
PreResNet-110	22.67	1,736,372	255.68M	Converted from GL model (log)
PreResNet-164(BN)	20.18	1,726,388	255.10M	Converted from GL model (log)
ResNeXt-29 (32x4d)	19.50	4,868,004	780.64M	Converted from GL model (log)
PyramidNet-110 (a=48)	20.95	1,778,556	408.38M	Converted from GL model (log)
PyramidNet-110 (a=84)	18.87	3,913,536	778.16M	Converted from GL model (log)
DenseNet-40 (k=12)	24.90	622,360	210.82M	Converted from GL model (log)
DenseNet-BC-40 (k=12)	28.41	188,092	74.90M	Converted from GL model (log)
DenseNet-BC-40 (k=24)	22.67	714,196	293.11M	Converted from GL model (log)
DenseNet-BC-40 (k=36)	20.50	1,578,412	654.64M	Converted from GL model (log)
DenseNet-100 (k=12)	19.64	4,129,600	1,353.62M	Converted from GL model (log)
DenseNet-BC-100 (k=12)	21.19	800,032	298.48M	Converted from GL model (log)
X-DenseNet-BC-40-2 (k=24)	23.96	714,196	293.11M	Converted from GL model (log)
X-DenseNet-BC-40-2 (k=36)	21.65	1,578,412	654.64M	Converted from GL model (log)
WRN-16-10	18.95	17,174,324	2,414.09M	Converted from GL model (log)
RoR-3-56	25.49	768,596	113.43M	Converted from GL model (log)
RoR-3-110	23.64	1,643,540	242.08M	Converted from GL model (log)
Shake-Shake-ResNet-20-2x16d	29.22	546,932	81.79M	Converted from GL model (log)
Shake-Shake-ResNet-26-2x32d	18.80	2,934,772	428.90M	Converted from GL model (log)

SVHN

Model	Error, %	Params	FLOPs/2	Remarks
ResNet-20	3.43	272,474	41.29M	Converted from GL model (log)
ResNet-56	2.75	855,770	127.06M	Converted from GL model (log)
ResNet-110	2.45	1,730,714	255.70M	Converted from GL model (log)
ResNet-164(BN)	2.42	1,704,154	255.31M	Converted from GL model (log)