DL4JSqueezeNet
Back to Model Zoo
======================================================================================================================================================
VertexName (VertexType) nIn,nOut TotalParams ParamsShape Vertex Inputs
======================================================================================================================================================
input_5 (InputVertex) -,- - - -
conv1 (ConvolutionLayer) 3,64 1,792 W:{64,3,3,3}, b:{1,64} [input_5]
relu1 (ActivationLayer) -,- 0 - [conv1]
pool1 (SubsamplingLayer) -,- 0 - [relu1]
fire2_squeeze1x1 (ConvolutionLayer) 64,16 1,040 W:{16,64,1,1}, b:{1,16} [pool1]
fire2_relu_squeeze1x1 (ActivationLayer) -,- 0 - [fire2_squeeze1x1]
fire2_expand1x1 (ConvolutionLayer) 16,64 1,088 W:{64,16,1,1}, b:{1,64} [fire2_relu_squeeze1x1]
fire2_expand3x3 (ConvolutionLayer) 16,64 9,280 W:{64,16,3,3}, b:{1,64} [fire2_relu_squeeze1x1]
fire2_relu_expand1x1 (ActivationLayer) -,- 0 - [fire2_expand1x1]
fire2_relu_expand3x3 (ActivationLayer) -,- 0 - [fire2_expand3x3]
fire2_concat (MergeVertex) -,- - - [fire2_relu_expand1x1, fire2_relu_expand3x3]
fire3_squeeze1x1 (ConvolutionLayer) 128,16 2,064 W:{16,128,1,1}, b:{1,16} [fire2_concat]
fire3_relu_squeeze1x1 (ActivationLayer) -,- 0 - [fire3_squeeze1x1]
fire3_expand1x1 (ConvolutionLayer) 16,64 1,088 W:{64,16,1,1}, b:{1,64} [fire3_relu_squeeze1x1]
fire3_expand3x3 (ConvolutionLayer) 16,64 9,280 W:{64,16,3,3}, b:{1,64} [fire3_relu_squeeze1x1]
fire3_relu_expand1x1 (ActivationLayer) -,- 0 - [fire3_expand1x1]
fire3_relu_expand3x3 (ActivationLayer) -,- 0 - [fire3_expand3x3]
fire3_concat (MergeVertex) -,- - - [fire3_relu_expand1x1, fire3_relu_expand3x3]
pool3 (SubsamplingLayer) -,- 0 - [fire3_concat]
fire4_squeeze1x1 (ConvolutionLayer) 128,32 4,128 W:{32,128,1,1}, b:{1,32} [pool3]
fire4_relu_squeeze1x1 (ActivationLayer) -,- 0 - [fire4_squeeze1x1]
fire4_expand1x1 (ConvolutionLayer) 32,128 4,224 W:{128,32,1,1}, b:{1,128} [fire4_relu_squeeze1x1]
fire4_expand3x3 (ConvolutionLayer) 32,128 36,992 W:{128,32,3,3}, b:{1,128} [fire4_relu_squeeze1x1]
fire4_relu_expand1x1 (ActivationLayer) -,- 0 - [fire4_expand1x1]
fire4_relu_expand3x3 (ActivationLayer) -,- 0 - [fire4_expand3x3]
fire4_concat (MergeVertex) -,- - - [fire4_relu_expand1x1, fire4_relu_expand3x3]
fire5_squeeze1x1 (ConvolutionLayer) 256,32 8,224 W:{32,256,1,1}, b:{1,32} [fire4_concat]
fire5_relu_squeeze1x1 (ActivationLayer) -,- 0 - [fire5_squeeze1x1]
fire5_expand1x1 (ConvolutionLayer) 32,128 4,224 W:{128,32,1,1}, b:{1,128} [fire5_relu_squeeze1x1]
fire5_expand3x3 (ConvolutionLayer) 32,128 36,992 W:{128,32,3,3}, b:{1,128} [fire5_relu_squeeze1x1]
fire5_relu_expand1x1 (ActivationLayer) -,- 0 - [fire5_expand1x1]
fire5_relu_expand3x3 (ActivationLayer) -,- 0 - [fire5_expand3x3]
fire5_concat (MergeVertex) -,- - - [fire5_relu_expand1x1, fire5_relu_expand3x3]
pool5 (SubsamplingLayer) -,- 0 - [fire5_concat]
fire6_squeeze1x1 (ConvolutionLayer) 256,48 12,336 W:{48,256,1,1}, b:{1,48} [pool5]
fire6_relu_squeeze1x1 (ActivationLayer) -,- 0 - [fire6_squeeze1x1]
fire6_expand1x1 (ConvolutionLayer) 48,192 9,408 W:{192,48,1,1}, b:{1,192} [fire6_relu_squeeze1x1]
fire6_expand3x3 (ConvolutionLayer) 48,192 83,136 W:{192,48,3,3}, b:{1,192} [fire6_relu_squeeze1x1]
fire6_relu_expand1x1 (ActivationLayer) -,- 0 - [fire6_expand1x1]
fire6_relu_expand3x3 (ActivationLayer) -,- 0 - [fire6_expand3x3]
fire6_concat (MergeVertex) -,- - - [fire6_relu_expand1x1, fire6_relu_expand3x3]
fire7_squeeze1x1 (ConvolutionLayer) 384,48 18,480 W:{48,384,1,1}, b:{1,48} [fire6_concat]
fire7_relu_squeeze1x1 (ActivationLayer) -,- 0 - [fire7_squeeze1x1]
fire7_expand1x1 (ConvolutionLayer) 48,192 9,408 W:{192,48,1,1}, b:{1,192} [fire7_relu_squeeze1x1]
fire7_expand3x3 (ConvolutionLayer) 48,192 83,136 W:{192,48,3,3}, b:{1,192} [fire7_relu_squeeze1x1]
fire7_relu_expand1x1 (ActivationLayer) -,- 0 - [fire7_expand1x1]
fire7_relu_expand3x3 (ActivationLayer) -,- 0 - [fire7_expand3x3]
fire7_concat (MergeVertex) -,- - - [fire7_relu_expand1x1, fire7_relu_expand3x3]
fire8_squeeze1x1 (ConvolutionLayer) 384,64 24,640 W:{64,384,1,1}, b:{1,64} [fire7_concat]
fire8_relu_squeeze1x1 (ActivationLayer) -,- 0 - [fire8_squeeze1x1]
fire8_expand1x1 (ConvolutionLayer) 64,256 16,640 W:{256,64,1,1}, b:{1,256} [fire8_relu_squeeze1x1]
fire8_expand3x3 (ConvolutionLayer) 64,256 147,712 W:{256,64,3,3}, b:{1,256} [fire8_relu_squeeze1x1]
fire8_relu_expand1x1 (ActivationLayer) -,- 0 - [fire8_expand1x1]
fire8_relu_expand3x3 (ActivationLayer) -,- 0 - [fire8_expand3x3]
fire8_concat (MergeVertex) -,- - - [fire8_relu_expand1x1, fire8_relu_expand3x3]
fire9_squeeze1x1 (ConvolutionLayer) 512,64 32,832 W:{64,512,1,1}, b:{1,64} [fire8_concat]
fire9_relu_squeeze1x1 (ActivationLayer) -,- 0 - [fire9_squeeze1x1]
fire9_expand1x1 (ConvolutionLayer) 64,256 16,640 W:{256,64,1,1}, b:{1,256} [fire9_relu_squeeze1x1]
fire9_expand3x3 (ConvolutionLayer) 64,256 147,712 W:{256,64,3,3}, b:{1,256} [fire9_relu_squeeze1x1]
fire9_relu_expand1x1 (ActivationLayer) -,- 0 - [fire9_expand1x1]
fire9_relu_expand3x3 (ActivationLayer) -,- 0 - [fire9_expand3x3]
fire9_concat (MergeVertex) -,- - - [fire9_relu_expand1x1, fire9_relu_expand3x3]
drop9 (DropoutLayer) -,- 0 - [fire9_concat]
conv10 (ConvolutionLayer) 512,1000 513,000 W:{1000,512,1,1}, b:{1,1000} [drop9]
relu10 (ActivationLayer) -,- 0 - [conv10]
global_average_pooling2d_5 (GlobalPoolingLayer) -,- 0 - [relu10]
loss (ActivationLayer) -,- 0 - [global_average_pooling2d_5]
------------------------------------------------------------------------------------------------------------------------------------------------------
Total Parameters: 1,235,496
Trainable Parameters: 1,235,496
Frozen Parameters: 0
======================================================================================================================================================