I did everything as per the assignments instructions with a dropout layer and image augmentation and the checkpoints return matching output; except the summary for just the pretrained InceptionV3 model looked slightly different; I assumed it was fine since the connected model has my dense layers connected to mixed7 with the dropout. The code only runs for one epoch (with no steps) and achieves 100% accuracy on training while only 50% on validation. I don’t understand the issue because everything seems right.
Heres the final model summary:
Model: “model_3”
Layer (type) Output Shape Param # Connected to
input_1 (InputLayer) [(None, 150, 150, 3)] 0
conv2d (Conv2D) (None, 74, 74, 32) 864 [‘input_1[0][0]’]
batch_normalization (Batch (None, 74, 74, 32) 96 [‘conv2d[0][0]’]
Normalization)
activation (Activation) (None, 74, 74, 32) 0 [‘batch_normalization[0][0]’]
conv2d_1 (Conv2D) (None, 72, 72, 32) 9216 [‘activation[0][0]’]
batch_normalization_1 (Bat (None, 72, 72, 32) 96 [‘conv2d_1[0][0]’]
chNormalization)
activation_1 (Activation) (None, 72, 72, 32) 0 [‘batch_normalization_1[0][0]’
]
conv2d_2 (Conv2D) (None, 72, 72, 64) 18432 [‘activation_1[0][0]’]
batch_normalization_2 (Bat (None, 72, 72, 64) 192 [‘conv2d_2[0][0]’]
chNormalization)
activation_2 (Activation) (None, 72, 72, 64) 0 [‘batch_normalization_2[0][0]’
]
max_pooling2d (MaxPooling2 (None, 35, 35, 64) 0 [‘activation_2[0][0]’]
D)
conv2d_3 (Conv2D) (None, 35, 35, 80) 5120 [‘max_pooling2d[0][0]’]
batch_normalization_3 (Bat (None, 35, 35, 80) 240 [‘conv2d_3[0][0]’]
chNormalization)
activation_3 (Activation) (None, 35, 35, 80) 0 [‘batch_normalization_3[0][0]’
]
conv2d_4 (Conv2D) (None, 33, 33, 192) 138240 [‘activation_3[0][0]’]
batch_normalization_4 (Bat (None, 33, 33, 192) 576 [‘conv2d_4[0][0]’]
chNormalization)
activation_4 (Activation) (None, 33, 33, 192) 0 [‘batch_normalization_4[0][0]’
]
max_pooling2d_1 (MaxPoolin (None, 16, 16, 192) 0 [‘activation_4[0][0]’]
g2D)
conv2d_8 (Conv2D) (None, 16, 16, 64) 12288 [‘max_pooling2d_1[0][0]’]
batch_normalization_8 (Bat (None, 16, 16, 64) 192 [‘conv2d_8[0][0]’]
chNormalization)
activation_8 (Activation) (None, 16, 16, 64) 0 [‘batch_normalization_8[0][0]’
]
conv2d_6 (Conv2D) (None, 16, 16, 48) 9216 [‘max_pooling2d_1[0][0]’]
conv2d_9 (Conv2D) (None, 16, 16, 96) 55296 [‘activation_8[0][0]’]
batch_normalization_6 (Bat (None, 16, 16, 48) 144 [‘conv2d_6[0][0]’]
chNormalization)
batch_normalization_9 (Bat (None, 16, 16, 96) 288 [‘conv2d_9[0][0]’]
chNormalization)
activation_6 (Activation) (None, 16, 16, 48) 0 [‘batch_normalization_6[0][0]’
]
activation_9 (Activation) (None, 16, 16, 96) 0 [‘batch_normalization_9[0][0]’
]
average_pooling2d (Average (None, 16, 16, 192) 0 [‘max_pooling2d_1[0][0]’]
Pooling2D)
conv2d_5 (Conv2D) (None, 16, 16, 64) 12288 [‘max_pooling2d_1[0][0]’]
conv2d_7 (Conv2D) (None, 16, 16, 64) 76800 [‘activation_6[0][0]’]
conv2d_10 (Conv2D) (None, 16, 16, 96) 82944 [‘activation_9[0][0]’]
conv2d_11 (Conv2D) (None, 16, 16, 32) 6144 [‘average_pooling2d[0][0]’]
batch_normalization_5 (Bat (None, 16, 16, 64) 192 [‘conv2d_5[0][0]’]
chNormalization)
batch_normalization_7 (Bat (None, 16, 16, 64) 192 [‘conv2d_7[0][0]’]
chNormalization)
batch_normalization_10 (Ba (None, 16, 16, 96) 288 [‘conv2d_10[0][0]’]
tchNormalization)
batch_normalization_11 (Ba (None, 16, 16, 32) 96 [‘conv2d_11[0][0]’]
tchNormalization)
activation_5 (Activation) (None, 16, 16, 64) 0 [‘batch_normalization_5[0][0]’
]
activation_7 (Activation) (None, 16, 16, 64) 0 [‘batch_normalization_7[0][0]’
]
activation_10 (Activation) (None, 16, 16, 96) 0 ['batch_normalization_10[0][0]
']
activation_11 (Activation) (None, 16, 16, 32) 0 ['batch_normalization_11[0][0]
']
mixed0 (Concatenate) (None, 16, 16, 256) 0 [‘activation_5[0][0]’,
‘activation_7[0][0]’,
‘activation_10[0][0]’,
‘activation_11[0][0]’]
conv2d_15 (Conv2D) (None, 16, 16, 64) 16384 [‘mixed0[0][0]’]
batch_normalization_15 (Ba (None, 16, 16, 64) 192 [‘conv2d_15[0][0]’]
tchNormalization)
activation_15 (Activation) (None, 16, 16, 64) 0 ['batch_normalization_15[0][0]
']
conv2d_13 (Conv2D) (None, 16, 16, 48) 12288 [‘mixed0[0][0]’]
conv2d_16 (Conv2D) (None, 16, 16, 96) 55296 [‘activation_15[0][0]’]
batch_normalization_13 (Ba (None, 16, 16, 48) 144 [‘conv2d_13[0][0]’]
tchNormalization)
batch_normalization_16 (Ba (None, 16, 16, 96) 288 [‘conv2d_16[0][0]’]
tchNormalization)
activation_13 (Activation) (None, 16, 16, 48) 0 ['batch_normalization_13[0][0]
']
activation_16 (Activation) (None, 16, 16, 96) 0 ['batch_normalization_16[0][0]
']
average_pooling2d_1 (Avera (None, 16, 16, 256) 0 [‘mixed0[0][0]’]
gePooling2D)
conv2d_12 (Conv2D) (None, 16, 16, 64) 16384 [‘mixed0[0][0]’]
conv2d_14 (Conv2D) (None, 16, 16, 64) 76800 [‘activation_13[0][0]’]
conv2d_17 (Conv2D) (None, 16, 16, 96) 82944 [‘activation_16[0][0]’]
conv2d_18 (Conv2D) (None, 16, 16, 64) 16384 [‘average_pooling2d_1[0][0]’]
batch_normalization_12 (Ba (None, 16, 16, 64) 192 [‘conv2d_12[0][0]’]
tchNormalization)
batch_normalization_14 (Ba (None, 16, 16, 64) 192 [‘conv2d_14[0][0]’]
tchNormalization)
batch_normalization_17 (Ba (None, 16, 16, 96) 288 [‘conv2d_17[0][0]’]
tchNormalization)
batch_normalization_18 (Ba (None, 16, 16, 64) 192 [‘conv2d_18[0][0]’]
tchNormalization)
activation_12 (Activation) (None, 16, 16, 64) 0 ['batch_normalization_12[0][0]
']
activation_14 (Activation) (None, 16, 16, 64) 0 ['batch_normalization_14[0][0]
']
activation_17 (Activation) (None, 16, 16, 96) 0 ['batch_normalization_17[0][0]
']
activation_18 (Activation) (None, 16, 16, 64) 0 ['batch_normalization_18[0][0]
']
mixed1 (Concatenate) (None, 16, 16, 288) 0 [‘activation_12[0][0]’,
‘activation_14[0][0]’,
‘activation_17[0][0]’,
‘activation_18[0][0]’]
conv2d_22 (Conv2D) (None, 16, 16, 64) 18432 [‘mixed1[0][0]’]
batch_normalization_22 (Ba (None, 16, 16, 64) 192 [‘conv2d_22[0][0]’]
tchNormalization)
activation_22 (Activation) (None, 16, 16, 64) 0 ['batch_normalization_22[0][0]
']
conv2d_20 (Conv2D) (None, 16, 16, 48) 13824 [‘mixed1[0][0]’]
conv2d_23 (Conv2D) (None, 16, 16, 96) 55296 [‘activation_22[0][0]’]
batch_normalization_20 (Ba (None, 16, 16, 48) 144 [‘conv2d_20[0][0]’]
tchNormalization)
batch_normalization_23 (Ba (None, 16, 16, 96) 288 [‘conv2d_23[0][0]’]
tchNormalization)
activation_20 (Activation) (None, 16, 16, 48) 0 ['batch_normalization_20[0][0]
']
activation_23 (Activation) (None, 16, 16, 96) 0 ['batch_normalization_23[0][0]
']
average_pooling2d_2 (Avera (None, 16, 16, 288) 0 [‘mixed1[0][0]’]
gePooling2D)
conv2d_19 (Conv2D) (None, 16, 16, 64) 18432 [‘mixed1[0][0]’]
conv2d_21 (Conv2D) (None, 16, 16, 64) 76800 [‘activation_20[0][0]’]
conv2d_24 (Conv2D) (None, 16, 16, 96) 82944 [‘activation_23[0][0]’]
conv2d_25 (Conv2D) (None, 16, 16, 64) 18432 [‘average_pooling2d_2[0][0]’]
batch_normalization_19 (Ba (None, 16, 16, 64) 192 [‘conv2d_19[0][0]’]
tchNormalization)
batch_normalization_21 (Ba (None, 16, 16, 64) 192 [‘conv2d_21[0][0]’]
tchNormalization)
batch_normalization_24 (Ba (None, 16, 16, 96) 288 [‘conv2d_24[0][0]’]
tchNormalization)
batch_normalization_25 (Ba (None, 16, 16, 64) 192 [‘conv2d_25[0][0]’]
tchNormalization)
activation_19 (Activation) (None, 16, 16, 64) 0 ['batch_normalization_19[0][0]
']
activation_21 (Activation) (None, 16, 16, 64) 0 ['batch_normalization_21[0][0]
']
activation_24 (Activation) (None, 16, 16, 96) 0 ['batch_normalization_24[0][0]
']
activation_25 (Activation) (None, 16, 16, 64) 0 ['batch_normalization_25[0][0]
']
mixed2 (Concatenate) (None, 16, 16, 288) 0 [‘activation_19[0][0]’,
‘activation_21[0][0]’,
‘activation_24[0][0]’,
‘activation_25[0][0]’]
conv2d_27 (Conv2D) (None, 16, 16, 64) 18432 [‘mixed2[0][0]’]
batch_normalization_27 (Ba (None, 16, 16, 64) 192 [‘conv2d_27[0][0]’]
tchNormalization)
activation_27 (Activation) (None, 16, 16, 64) 0 ['batch_normalization_27[0][0]
']
conv2d_28 (Conv2D) (None, 16, 16, 96) 55296 [‘activation_27[0][0]’]
batch_normalization_28 (Ba (None, 16, 16, 96) 288 [‘conv2d_28[0][0]’]
tchNormalization)
activation_28 (Activation) (None, 16, 16, 96) 0 ['batch_normalization_28[0][0]
']
conv2d_26 (Conv2D) (None, 7, 7, 384) 995328 [‘mixed2[0][0]’]
conv2d_29 (Conv2D) (None, 7, 7, 96) 82944 [‘activation_28[0][0]’]
batch_normalization_26 (Ba (None, 7, 7, 384) 1152 [‘conv2d_26[0][0]’]
tchNormalization)
batch_normalization_29 (Ba (None, 7, 7, 96) 288 [‘conv2d_29[0][0]’]
tchNormalization)
activation_26 (Activation) (None, 7, 7, 384) 0 ['batch_normalization_26[0][0]
']
activation_29 (Activation) (None, 7, 7, 96) 0 ['batch_normalization_29[0][0]
']
max_pooling2d_2 (MaxPoolin (None, 7, 7, 288) 0 [‘mixed2[0][0]’]
g2D)
mixed3 (Concatenate) (None, 7, 7, 768) 0 [‘activation_26[0][0]’,
‘activation_29[0][0]’,
‘max_pooling2d_2[0][0]’]
conv2d_34 (Conv2D) (None, 7, 7, 128) 98304 [‘mixed3[0][0]’]
batch_normalization_34 (Ba (None, 7, 7, 128) 384 [‘conv2d_34[0][0]’]
tchNormalization)
activation_34 (Activation) (None, 7, 7, 128) 0 ['batch_normalization_34[0][0]
']
conv2d_35 (Conv2D) (None, 7, 7, 128) 114688 [‘activation_34[0][0]’]
batch_normalization_35 (Ba (None, 7, 7, 128) 384 [‘conv2d_35[0][0]’]
tchNormalization)
activation_35 (Activation) (None, 7, 7, 128) 0 ['batch_normalization_35[0][0]
']
conv2d_31 (Conv2D) (None, 7, 7, 128) 98304 [‘mixed3[0][0]’]
conv2d_36 (Conv2D) (None, 7, 7, 128) 114688 [‘activation_35[0][0]’]
batch_normalization_31 (Ba (None, 7, 7, 128) 384 [‘conv2d_31[0][0]’]
tchNormalization)
batch_normalization_36 (Ba (None, 7, 7, 128) 384 [‘conv2d_36[0][0]’]
tchNormalization)
activation_31 (Activation) (None, 7, 7, 128) 0 ['batch_normalization_31[0][0]
']
activation_36 (Activation) (None, 7, 7, 128) 0 ['batch_normalization_36[0][0]
']
conv2d_32 (Conv2D) (None, 7, 7, 128) 114688 [‘activation_31[0][0]’]
conv2d_37 (Conv2D) (None, 7, 7, 128) 114688 [‘activation_36[0][0]’]
batch_normalization_32 (Ba (None, 7, 7, 128) 384 [‘conv2d_32[0][0]’]
tchNormalization)
batch_normalization_37 (Ba (None, 7, 7, 128) 384 [‘conv2d_37[0][0]’]
tchNormalization)
activation_32 (Activation) (None, 7, 7, 128) 0 ['batch_normalization_32[0][0]
']
activation_37 (Activation) (None, 7, 7, 128) 0 ['batch_normalization_37[0][0]
']
average_pooling2d_3 (Avera (None, 7, 7, 768) 0 [‘mixed3[0][0]’]
gePooling2D)
conv2d_30 (Conv2D) (None, 7, 7, 192) 147456 [‘mixed3[0][0]’]
conv2d_33 (Conv2D) (None, 7, 7, 192) 172032 [‘activation_32[0][0]’]
conv2d_38 (Conv2D) (None, 7, 7, 192) 172032 [‘activation_37[0][0]’]
conv2d_39 (Conv2D) (None, 7, 7, 192) 147456 [‘average_pooling2d_3[0][0]’]
batch_normalization_30 (Ba (None, 7, 7, 192) 576 [‘conv2d_30[0][0]’]
tchNormalization)
batch_normalization_33 (Ba (None, 7, 7, 192) 576 [‘conv2d_33[0][0]’]
tchNormalization)
batch_normalization_38 (Ba (None, 7, 7, 192) 576 [‘conv2d_38[0][0]’]
tchNormalization)
batch_normalization_39 (Ba (None, 7, 7, 192) 576 [‘conv2d_39[0][0]’]
tchNormalization)
activation_30 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_30[0][0]
']
activation_33 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_33[0][0]
']
activation_38 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_38[0][0]
']
activation_39 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_39[0][0]
']
mixed4 (Concatenate) (None, 7, 7, 768) 0 [‘activation_30[0][0]’,
‘activation_33[0][0]’,
‘activation_38[0][0]’,
‘activation_39[0][0]’]
conv2d_44 (Conv2D) (None, 7, 7, 160) 122880 [‘mixed4[0][0]’]
batch_normalization_44 (Ba (None, 7, 7, 160) 480 [‘conv2d_44[0][0]’]
tchNormalization)
activation_44 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_44[0][0]
']
conv2d_45 (Conv2D) (None, 7, 7, 160) 179200 [‘activation_44[0][0]’]
batch_normalization_45 (Ba (None, 7, 7, 160) 480 [‘conv2d_45[0][0]’]
tchNormalization)
activation_45 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_45[0][0]
']
conv2d_41 (Conv2D) (None, 7, 7, 160) 122880 [‘mixed4[0][0]’]
conv2d_46 (Conv2D) (None, 7, 7, 160) 179200 [‘activation_45[0][0]’]
batch_normalization_41 (Ba (None, 7, 7, 160) 480 [‘conv2d_41[0][0]’]
tchNormalization)
batch_normalization_46 (Ba (None, 7, 7, 160) 480 [‘conv2d_46[0][0]’]
tchNormalization)
activation_41 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_41[0][0]
']
activation_46 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_46[0][0]
']
conv2d_42 (Conv2D) (None, 7, 7, 160) 179200 [‘activation_41[0][0]’]
conv2d_47 (Conv2D) (None, 7, 7, 160) 179200 [‘activation_46[0][0]’]
batch_normalization_42 (Ba (None, 7, 7, 160) 480 [‘conv2d_42[0][0]’]
tchNormalization)
batch_normalization_47 (Ba (None, 7, 7, 160) 480 [‘conv2d_47[0][0]’]
tchNormalization)
activation_42 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_42[0][0]
']
activation_47 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_47[0][0]
']
average_pooling2d_4 (Avera (None, 7, 7, 768) 0 [‘mixed4[0][0]’]
gePooling2D)
conv2d_40 (Conv2D) (None, 7, 7, 192) 147456 [‘mixed4[0][0]’]
conv2d_43 (Conv2D) (None, 7, 7, 192) 215040 [‘activation_42[0][0]’]
conv2d_48 (Conv2D) (None, 7, 7, 192) 215040 [‘activation_47[0][0]’]
conv2d_49 (Conv2D) (None, 7, 7, 192) 147456 [‘average_pooling2d_4[0][0]’]
batch_normalization_40 (Ba (None, 7, 7, 192) 576 [‘conv2d_40[0][0]’]
tchNormalization)
batch_normalization_43 (Ba (None, 7, 7, 192) 576 [‘conv2d_43[0][0]’]
tchNormalization)
batch_normalization_48 (Ba (None, 7, 7, 192) 576 [‘conv2d_48[0][0]’]
tchNormalization)
batch_normalization_49 (Ba (None, 7, 7, 192) 576 [‘conv2d_49[0][0]’]
tchNormalization)
activation_40 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_40[0][0]
']
activation_43 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_43[0][0]
']
activation_48 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_48[0][0]
']
activation_49 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_49[0][0]
']
mixed5 (Concatenate) (None, 7, 7, 768) 0 [‘activation_40[0][0]’,
‘activation_43[0][0]’,
‘activation_48[0][0]’,
‘activation_49[0][0]’]
conv2d_54 (Conv2D) (None, 7, 7, 160) 122880 [‘mixed5[0][0]’]
batch_normalization_54 (Ba (None, 7, 7, 160) 480 [‘conv2d_54[0][0]’]
tchNormalization)
activation_54 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_54[0][0]
']
conv2d_55 (Conv2D) (None, 7, 7, 160) 179200 [‘activation_54[0][0]’]
batch_normalization_55 (Ba (None, 7, 7, 160) 480 [‘conv2d_55[0][0]’]
tchNormalization)
activation_55 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_55[0][0]
']
conv2d_51 (Conv2D) (None, 7, 7, 160) 122880 [‘mixed5[0][0]’]
conv2d_56 (Conv2D) (None, 7, 7, 160) 179200 [‘activation_55[0][0]’]
batch_normalization_51 (Ba (None, 7, 7, 160) 480 [‘conv2d_51[0][0]’]
tchNormalization)
batch_normalization_56 (Ba (None, 7, 7, 160) 480 [‘conv2d_56[0][0]’]
tchNormalization)
activation_51 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_51[0][0]
']
activation_56 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_56[0][0]
']
conv2d_52 (Conv2D) (None, 7, 7, 160) 179200 [‘activation_51[0][0]’]
conv2d_57 (Conv2D) (None, 7, 7, 160) 179200 [‘activation_56[0][0]’]
batch_normalization_52 (Ba (None, 7, 7, 160) 480 [‘conv2d_52[0][0]’]
tchNormalization)
batch_normalization_57 (Ba (None, 7, 7, 160) 480 [‘conv2d_57[0][0]’]
tchNormalization)
activation_52 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_52[0][0]
']
activation_57 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_57[0][0]
']
average_pooling2d_5 (Avera (None, 7, 7, 768) 0 [‘mixed5[0][0]’]
gePooling2D)
conv2d_50 (Conv2D) (None, 7, 7, 192) 147456 [‘mixed5[0][0]’]
conv2d_53 (Conv2D) (None, 7, 7, 192) 215040 [‘activation_52[0][0]’]
conv2d_58 (Conv2D) (None, 7, 7, 192) 215040 [‘activation_57[0][0]’]
conv2d_59 (Conv2D) (None, 7, 7, 192) 147456 [‘average_pooling2d_5[0][0]’]
batch_normalization_50 (Ba (None, 7, 7, 192) 576 [‘conv2d_50[0][0]’]
tchNormalization)
batch_normalization_53 (Ba (None, 7, 7, 192) 576 [‘conv2d_53[0][0]’]
tchNormalization)
batch_normalization_58 (Ba (None, 7, 7, 192) 576 [‘conv2d_58[0][0]’]
tchNormalization)
batch_normalization_59 (Ba (None, 7, 7, 192) 576 [‘conv2d_59[0][0]’]
tchNormalization)
activation_50 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_50[0][0]
']
activation_53 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_53[0][0]
']
activation_58 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_58[0][0]
']
activation_59 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_59[0][0]
']
mixed6 (Concatenate) (None, 7, 7, 768) 0 [‘activation_50[0][0]’,
‘activation_53[0][0]’,
‘activation_58[0][0]’,
‘activation_59[0][0]’]
conv2d_64 (Conv2D) (None, 7, 7, 192) 147456 [‘mixed6[0][0]’]
batch_normalization_64 (Ba (None, 7, 7, 192) 576 [‘conv2d_64[0][0]’]
tchNormalization)
activation_64 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_64[0][0]
']
conv2d_65 (Conv2D) (None, 7, 7, 192) 258048 [‘activation_64[0][0]’]
batch_normalization_65 (Ba (None, 7, 7, 192) 576 [‘conv2d_65[0][0]’]
tchNormalization)
activation_65 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_65[0][0]
']
conv2d_61 (Conv2D) (None, 7, 7, 192) 147456 [‘mixed6[0][0]’]
conv2d_66 (Conv2D) (None, 7, 7, 192) 258048 [‘activation_65[0][0]’]
batch_normalization_61 (Ba (None, 7, 7, 192) 576 [‘conv2d_61[0][0]’]
tchNormalization)
batch_normalization_66 (Ba (None, 7, 7, 192) 576 [‘conv2d_66[0][0]’]
tchNormalization)
activation_61 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_61[0][0]
']
activation_66 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_66[0][0]
']
conv2d_62 (Conv2D) (None, 7, 7, 192) 258048 [‘activation_61[0][0]’]
conv2d_67 (Conv2D) (None, 7, 7, 192) 258048 [‘activation_66[0][0]’]
batch_normalization_62 (Ba (None, 7, 7, 192) 576 [‘conv2d_62[0][0]’]
tchNormalization)
batch_normalization_67 (Ba (None, 7, 7, 192) 576 [‘conv2d_67[0][0]’]
tchNormalization)
activation_62 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_62[0][0]
']
activation_67 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_67[0][0]
']
average_pooling2d_6 (Avera (None, 7, 7, 768) 0 [‘mixed6[0][0]’]
gePooling2D)
conv2d_60 (Conv2D) (None, 7, 7, 192) 147456 [‘mixed6[0][0]’]
conv2d_63 (Conv2D) (None, 7, 7, 192) 258048 [‘activation_62[0][0]’]
conv2d_68 (Conv2D) (None, 7, 7, 192) 258048 [‘activation_67[0][0]’]
conv2d_69 (Conv2D) (None, 7, 7, 192) 147456 [‘average_pooling2d_6[0][0]’]
batch_normalization_60 (Ba (None, 7, 7, 192) 576 [‘conv2d_60[0][0]’]
tchNormalization)
batch_normalization_63 (Ba (None, 7, 7, 192) 576 [‘conv2d_63[0][0]’]
tchNormalization)
batch_normalization_68 (Ba (None, 7, 7, 192) 576 [‘conv2d_68[0][0]’]
tchNormalization)
batch_normalization_69 (Ba (None, 7, 7, 192) 576 [‘conv2d_69[0][0]’]
tchNormalization)
activation_60 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_60[0][0]
']
activation_63 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_63[0][0]
']
activation_68 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_68[0][0]
']
activation_69 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_69[0][0]
']
mixed7 (Concatenate) (None, 7, 7, 768) 0 [‘activation_60[0][0]’,
‘activation_63[0][0]’,
‘activation_68[0][0]’,
‘activation_69[0][0]’]
flatten_3 (Flatten) (None, 37632) 0 [‘mixed7[0][0]’]
dense_6 (Dense) (None, 1024) 3853619 [‘flatten_3[0][0]’]
2
dropout_3 (Dropout) (None, 1024) 0 [‘dense_6[0][0]’]
dense_7 (Dense) (None, 1) 1025 [‘dropout_3[0][0]’]
==================================================================================================
Total params: 47512481 (181.25 MB)
Trainable params: 38537217 (147.01 MB)
Non-trainable params: 8975264 (34.24 MB)
None
There are 47,512,481 total parameters in this model.
There are 38,537,217 trainable parameters in this model. (these last lines matched expected output).