diff --git a/mobilenet_v3_large_1.0.prototxt b/mobilenet_v3_large_1.0.prototxt
index b9a0f89..30af007 100644
--- a/mobilenet_v3_large_1.0.prototxt
+++ b/mobilenet_v3_large_1.0.prototxt
@@ -549,12 +549,6 @@ layer {
     bias_term: true
   }
 }
-layer {
-  name: "conv3-2/dwise-relu"
-  type: "ReLU6"
-  bottom: "conv3-2/dwise"
-  top: "conv3-2/dwise"
-}
 
 # se
 layer {
@@ -660,6 +654,12 @@ layer {
     axis: 0
   }
 }
+layer {
+  name: "conv3-2/dwise/scale-relu"
+  type: "ReLU6"
+  bottom: "conv3-2/dwise/scale"
+  top: "conv3-2/dwise/scale"
+}
 
 layer {
   name: "conv3-2/linear"
@@ -778,12 +778,6 @@ layer {
     bias_term: true
   }
 }
-layer {
-  name: "conv4-1/dwise-relu"
-  type: "ReLU6"
-  bottom: "conv4-1/dwise"
-  top: "conv4-1/dwise"
-}
 
 # se
 layer {
@@ -889,6 +883,12 @@ layer {
     axis: 0
   }
 }
+layer {
+  name: "conv4-1/dwise/scale-relu"
+  type: "ReLU6"
+  bottom: "conv4-1/dwise/scale"
+  top: "conv4-1/dwise/scale"
+}
 
 layer {
   name: "conv4-1/linear"
@@ -1015,12 +1015,6 @@ layer {
     bias_term: true
   }
 }
-layer {
-  name: "conv4-2/dwise-relu"
-  type: "ReLU6"
-  bottom: "conv4-2/dwise"
-  top: "conv4-2/dwise"
-}
 
 # se
 layer {
@@ -1126,6 +1120,12 @@ layer {
     axis: 0
   }
 }
+layer {
+  name: "conv4-2/dwise/scale-relu"
+  type: "ReLU6"
+  bottom: "conv4-2/dwise/scale"
+  top: "conv4-2/dwise/scale"
+}
 
 layer {
   name: "conv4-2/linear"
@@ -2069,51 +2069,12 @@ layer {
     bias_term: true
   }
 }
-# H-swish
-layer {
-  name: "conv5-4/dwise/shift"
-  type: "Power"
-  bottom: "conv5-4/dwise"
-  top: "conv5-4/dwise/shift"
-  power_param {
-    power: 1
-    scale: 1
-    shift: 3
-  }
-}
-layer {
-  name: "conv5-4/dwise/shift-relu"
-  type: "ReLU6"
-  bottom: "conv5-4/dwise/shift"
-  top: "conv5-4/dwise/shift"
-}
-layer {
-  name: "conv5-4/dwise/shift-div"
-  type: "Power"
-  bottom: "conv5-4/dwise/shift"
-  top: "conv5-4/dwise/shift-div"
-  power_param {
-    power: 1
-    scale: 0.1666666667
-    shift: 0
-  }
-}
-layer {
-  name: "conv5-4/dwise/hswish"
-  type: "Eltwise"
-  bottom: "conv5-4/dwise"
-  bottom: "conv5-4/dwise/shift-div"
-  top: "conv5-4/dwise/hswish"
-  eltwise_param {
-    operation: PROD
-  }
-}
 
 # se
 layer {
   name: "conv5-4/dwise-se-pool"
   type: "Pooling"
-  bottom: "conv5-4/dwise/hswish"
+  bottom: "conv5-4/dwise"
   top: "conv5-4/dwise-se-pool"
   pooling_param {
     pool: AVE
@@ -2206,18 +2167,57 @@ layer {
 layer {
   name: "conv5-4/dwise/scale"
   type: "Scale"
-  bottom: "conv5-4/dwise/hswish"
+  bottom: "conv5-4/dwise"
   bottom: "conv5-4/dwise-se-fc2/shift-div"
   top: "conv5-4/dwise/scale"
   scale_param{
     axis: 0
   }
 }
+# H-swish
+layer {
+  name: "conv5-4/dwise/shift"
+  type: "Power"
+  bottom: "conv5-4/dwise/scale"
+  top: "conv5-4/dwise/shift"
+  power_param {
+    power: 1
+    scale: 1
+    shift: 3
+  }
+}
+layer {
+  name: "conv5-4/dwise/shift-relu"
+  type: "ReLU6"
+  bottom: "conv5-4/dwise/shift"
+  top: "conv5-4/dwise/shift"
+}
+layer {
+  name: "conv5-4/dwise/shift-div"
+  type: "Power"
+  bottom: "conv5-4/dwise/shift"
+  top: "conv5-4/dwise/shift-div"
+  power_param {
+    power: 1
+    scale: 0.1666666667
+    shift: 0
+  }
+}
+layer {
+  name: "conv5-4/dwise/scale/hswish"
+  type: "Eltwise"
+  bottom: "conv5-4/dwise/scale"
+  bottom: "conv5-4/dwise/shift-div"
+  top: "conv5-4/dwise/scale/hswish"
+  eltwise_param {
+    operation: PROD
+  }
+}
 
 layer {
   name: "conv5-4/linear"
   type: "Convolution"
-  bottom: "conv5-4/dwise/scale"
+  bottom: "conv5-4/dwise/scale/hswish"
   top: "conv5-4/linear"
   param {
     lr_mult: 1
@@ -2364,51 +2364,12 @@ layer {
     bias_term: true
   }
 }
-# H-swish
-layer {
-  name: "conv5-5/dwise/shift"
-  type: "Power"
-  bottom: "conv5-5/dwise"
-  top: "conv5-5/dwise/shift"
-  power_param {
-    power: 1
-    scale: 1
-    shift: 3
-  }
-}
-layer {
-  name: "conv5-5/dwise/shift-relu"
-  type: "ReLU6"
-  bottom: "conv5-5/dwise/shift"
-  top: "conv5-5/dwise/shift"
-}
-layer {
-  name: "conv5-5/dwise/shift-div"
-  type: "Power"
-  bottom: "conv5-5/dwise/shift"
-  top: "conv5-5/dwise/shift-div"
-  power_param {
-    power: 1
-    scale: 0.1666666667
-    shift: 0
-  }
-}
-layer {
-  name: "conv5-5/dwise/hswish"
-  type: "Eltwise"
-  bottom: "conv5-5/dwise"
-  bottom: "conv5-5/dwise/shift-div"
-  top: "conv5-5/dwise/hswish"
-  eltwise_param {
-    operation: PROD
-  }
-}
 
 # se
 layer {
   name: "conv5-5/dwise-se-pool"
   type: "Pooling"
-  bottom: "conv5-5/dwise/hswish"
+  bottom: "conv5-5/dwise"
   top: "conv5-5/dwise-se-pool"
   pooling_param {
     pool: AVE
@@ -2501,18 +2462,57 @@ layer {
 layer {
   name: "conv5-5/dwise/scale"
   type: "Scale"
-  bottom: "conv5-5/dwise/hswish"
+  bottom: "conv5-5/dwise"
   bottom: "conv5-5/dwise-se-fc2/shift-div"
   top: "conv5-5/dwise/scale"
   scale_param{
     axis: 0
   }
 }
+# H-swish
+layer {
+  name: "conv5-5/dwise/shift"
+  type: "Power"
+  bottom: "conv5-5/dwise/scale"
+  top: "conv5-5/dwise/shift"
+  power_param {
+    power: 1
+    scale: 1
+    shift: 3
+  }
+}
+layer {
+  name: "conv5-5/dwise/shift-relu"
+  type: "ReLU6"
+  bottom: "conv5-5/dwise/shift"
+  top: "conv5-5/dwise/shift"
+}
+layer {
+  name: "conv5-5/dwise/shift-div"
+  type: "Power"
+  bottom: "conv5-5/dwise/shift"
+  top: "conv5-5/dwise/shift-div"
+  power_param {
+    power: 1
+    scale: 0.1666666667
+    shift: 0
+  }
+}
+layer {
+  name: "conv5-5/dwise/hswish"
+  type: "Eltwise"
+  bottom: "conv5-5/dwise/scale"
+  bottom: "conv5-5/dwise/shift-div"
+  top: "conv5-5/dwise/hswish"
+  eltwise_param {
+    operation: PROD
+  }
+}
 
 layer {
   name: "conv5-5/linear"
   type: "Convolution"
-  bottom: "conv5-5/dwise/scale"
+  bottom: "conv5-5/dwise/hswish"
   top: "conv5-5/linear"
   param {
     lr_mult: 1
@@ -2667,51 +2667,12 @@ layer {
     bias_term: true
   }
 }
-# H-swish
-layer {
-  name: "conv5-6/dwise/shift"
-  type: "Power"
-  bottom: "conv5-6/dwise"
-  top: "conv5-6/dwise/shift"
-  power_param {
-    power: 1
-    scale: 1
-    shift: 3
-  }
-}
-layer {
-  name: "conv5-6/dwise/shift-relu"
-  type: "ReLU6"
-  bottom: "conv5-6/dwise/shift"
-  top: "conv5-6/dwise/shift"
-}
-layer {
-  name: "conv5-6/dwise/shift-div"
-  type: "Power"
-  bottom: "conv5-6/dwise/shift"
-  top: "conv5-6/dwise/shift-div"
-  power_param {
-    power: 1
-    scale: 0.1666666667
-    shift: 0
-  }
-}
-layer {
-  name: "conv5-6/dwise/hswish"
-  type: "Eltwise"
-  bottom: "conv5-6/dwise"
-  bottom: "conv5-6/dwise/shift-div"
-  top: "conv5-6/dwise/hswish"
-  eltwise_param {
-    operation: PROD
-  }
-}
 
 # se
 layer {
   name: "conv5-6/dwise-se-pool"
   type: "Pooling"
-  bottom: "conv5-6/dwise/hswish"
+  bottom: "conv5-6/dwise"
   top: "conv5-6/dwise-se-pool"
   pooling_param {
     pool: AVE
@@ -2804,18 +2765,57 @@ layer {
 layer {
   name: "conv5-6/dwise/scale"
   type: "Scale"
-  bottom: "conv5-6/dwise/hswish"
+  bottom: "conv5-6/dwise"
   bottom: "conv5-6/dwise-se-fc2/shift-div"
   top: "conv5-6/dwise/scale"
   scale_param{
     axis: 0
   }
 }
+# H-swish
+layer {
+  name: "conv5-6/dwise/shift"
+  type: "Power"
+  bottom: "conv5-6/dwise/scale"
+  top: "conv5-6/dwise/shift"
+  power_param {
+    power: 1
+    scale: 1
+    shift: 3
+  }
+}
+layer {
+  name: "conv5-6/dwise/shift-relu"
+  type: "ReLU6"
+  bottom: "conv5-6/dwise/shift"
+  top: "conv5-6/dwise/shift"
+}
+layer {
+  name: "conv5-6/dwise/shift-div"
+  type: "Power"
+  bottom: "conv5-6/dwise/shift"
+  top: "conv5-6/dwise/shift-div"
+  power_param {
+    power: 1
+    scale: 0.1666666667
+    shift: 0
+  }
+}
+layer {
+  name: "conv5-6/dwise/hswish"
+  type: "Eltwise"
+  bottom: "conv5-6/dwise/scale"
+  bottom: "conv5-6/dwise/shift-div"
+  top: "conv5-6/dwise/hswish"
+  eltwise_param {
+    operation: PROD
+  }
+}
 
 layer {
   name: "conv5-6/linear"
   type: "Convolution"
-  bottom: "conv5-6/dwise/scale"
+  bottom: "conv5-6/dwise/hswish"
   top: "conv5-6/linear"
   param {
     lr_mult: 1
@@ -2970,51 +2970,12 @@ layer {
     bias_term: true
   }
 }
-# H-swish
-layer {
-  name: "conv5-7/dwise/shift"
-  type: "Power"
-  bottom: "conv5-7/dwise"
-  top: "conv5-7/dwise/shift"
-  power_param {
-    power: 1
-    scale: 1
-    shift: 3
-  }
-}
-layer {
-  name: "conv5-7/dwise/shift-relu"
-  type: "ReLU6"
-  bottom: "conv5-7/dwise/shift"
-  top: "conv5-7/dwise/shift"
-}
-layer {
-  name: "conv5-7/dwise/shift-div"
-  type: "Power"
-  bottom: "conv5-7/dwise/shift"
-  top: "conv5-7/dwise/shift-div"
-  power_param {
-    power: 1
-    scale: 0.1666666667
-    shift: 0
-  }
-}
-layer {
-  name: "conv5-7/dwise/hswish"
-  type: "Eltwise"
-  bottom: "conv5-7/dwise"
-  bottom: "conv5-7/dwise/shift-div"
-  top: "conv5-7/dwise/hswish"
-  eltwise_param {
-    operation: PROD
-  }
-}
 
 # se
 layer {
   name: "conv5-7/dwise-se-pool"
   type: "Pooling"
-  bottom: "conv5-7/dwise/hswish"
+  bottom: "conv5-7/dwise"
   top: "conv5-7/dwise-se-pool"
   pooling_param {
     pool: AVE
@@ -3107,18 +3068,57 @@ layer {
 layer {
   name: "conv5-7/dwise/scale"
   type: "Scale"
-  bottom: "conv5-7/dwise/hswish"
+  bottom: "conv5-7/dwise"
   bottom: "conv5-7/dwise-se-fc2/shift-div"
   top: "conv5-7/dwise/scale"
   scale_param{
     axis: 0
   }
 }
+# H-swish
+layer {
+  name: "conv5-7/dwise/shift"
+  type: "Power"
+  bottom: "conv5-7/dwise/scale"
+  top: "conv5-7/dwise/shift"
+  power_param {
+    power: 1
+    scale: 1
+    shift: 3
+  }
+}
+layer {
+  name: "conv5-7/dwise/shift-relu"
+  type: "ReLU6"
+  bottom: "conv5-7/dwise/shift"
+  top: "conv5-7/dwise/shift"
+}
+layer {
+  name: "conv5-7/dwise/shift-div"
+  type: "Power"
+  bottom: "conv5-7/dwise/shift"
+  top: "conv5-7/dwise/shift-div"
+  power_param {
+    power: 1
+    scale: 0.1666666667
+    shift: 0
+  }
+}
+layer {
+  name: "conv5-7/dwise/hswish"
+  type: "Eltwise"
+  bottom: "conv5-7/dwise/scale"
+  bottom: "conv5-7/dwise/shift-div"
+  top: "conv5-7/dwise/hswish"
+  eltwise_param {
+    operation: PROD
+  }
+}
 
 layer {
   name: "conv5-7/linear"
   type: "Convolution"
-  bottom: "conv5-7/dwise/scale"
+  bottom: "conv5-7/dwise/hswish"
   top: "conv5-7/linear"
   param {
     lr_mult: 1
@@ -3265,51 +3265,12 @@ layer {
     bias_term: true
   }
 }
-# H-swish
-layer {
-  name: "conv6-1/dwise/shift"
-  type: "Power"
-  bottom: "conv6-1/dwise"
-  top: "conv6-1/dwise/shift"
-  power_param {
-    power: 1
-    scale: 1
-    shift: 3
-  }
-}
-layer {
-  name: "conv6-1/dwise/shift-relu"
-  type: "ReLU6"
-  bottom: "conv6-1/dwise/shift"
-  top: "conv6-1/dwise/shift"
-}
-layer {
-  name: "conv6-1/dwise/shift-div"
-  type: "Power"
-  bottom: "conv6-1/dwise/shift"
-  top: "conv6-1/dwise/shift-div"
-  power_param {
-    power: 1
-    scale: 0.1666666667
-    shift: 0
-  }
-}
-layer {
-  name: "conv6-1/dwise/hswish"
-  type: "Eltwise"
-  bottom: "conv6-1/dwise"
-  bottom: "conv6-1/dwise/shift-div"
-  top: "conv6-1/dwise/hswish"
-  eltwise_param {
-    operation: PROD
-  }
-}
 
 # se
 layer {
   name: "conv6-1/dwise-se-pool"
   type: "Pooling"
-  bottom: "conv6-1/dwise/hswish"
+  bottom: "conv6-1/dwise"
   top: "conv6-1/dwise-se-pool"
   pooling_param {
     pool: AVE
@@ -3402,18 +3363,57 @@ layer {
 layer {
   name: "conv6-1/dwise/scale"
   type: "Scale"
-  bottom: "conv6-1/dwise/hswish"
+  bottom: "conv6-1/dwise"
   bottom: "conv6-1/dwise-se-fc2/shift-div"
   top: "conv6-1/dwise/scale"
   scale_param{
     axis: 0
   }
 }
+# H-swish
+layer {
+  name: "conv6-1/dwise/shift"
+  type: "Power"
+  bottom: "conv6-1/dwise/scale"
+  top: "conv6-1/dwise/shift"
+  power_param {
+    power: 1
+    scale: 1
+    shift: 3
+  }
+}
+layer {
+  name: "conv6-1/dwise/shift-relu"
+  type: "ReLU6"
+  bottom: "conv6-1/dwise/shift"
+  top: "conv6-1/dwise/shift"
+}
+layer {
+  name: "conv6-1/dwise/shift-div"
+  type: "Power"
+  bottom: "conv6-1/dwise/shift"
+  top: "conv6-1/dwise/shift-div"
+  power_param {
+    power: 1
+    scale: 0.1666666667
+    shift: 0
+  }
+}
+layer {
+  name: "conv6-1/dwise/hswish"
+  type: "Eltwise"
+  bottom: "conv6-1/dwise/scale"
+  bottom: "conv6-1/dwise/shift-div"
+  top: "conv6-1/dwise/hswish"
+  eltwise_param {
+    operation: PROD
+  }
+}
 
 layer {
   name: "conv6-1/linear"
   type: "Convolution"
-  bottom: "conv6-1/dwise/scale"
+  bottom: "conv6-1/dwise/hswish"
   top: "conv6-1/linear"
   param {
     lr_mult: 1
@@ -3591,15 +3591,22 @@ layer {
     lr_mult: 1
     decay_mult: 1
   }
+  param {
+    lr_mult: 2
+    decay_mult: 0
+  }
   convolution_param {
     num_output: 1280
     pad: 0
-    bias_term: false
     kernel_size: 1
     stride: 1
     weight_filler {
       type: "msra"
     }
+    bias_filler {
+      type: "constant"
+      value: 0
+    }
   }
 }
 # H-swish
@@ -3652,15 +3659,22 @@ layer {
     lr_mult: 1
     decay_mult: 1
   }
+  param {
+    lr_mult: 2
+    decay_mult: 0
+  }
   convolution_param {
     num_output: 1000
     pad: 0
-    bias_term: false
     kernel_size: 1
     stride: 1
     weight_filler {
       type: "msra"
     }
+    bias_filler {
+      type: "constant"
+      value: 0
+    }
   }
 }
 
diff --git a/mobilenet_v3_small_1.0.prototxt b/mobilenet_v3_small_1.0.prototxt
index ab57a83..e73dc2c 100644
--- a/mobilenet_v3_small_1.0.prototxt
+++ b/mobilenet_v3_small_1.0.prototxt
@@ -161,12 +161,6 @@ layer {
     bias_term: true
   }
 }
-layer {
-  name: "conv2-1/dwise-relu"
-  type: "ReLU6"
-  bottom: "conv2-1/dwise"
-  top: "conv2-1/dwise"
-}
 
 # se
 layer {
@@ -272,6 +266,12 @@ layer {
     axis: 0
   }
 }
+layer {
+  name: "conv2-1/dwise/scale-relu"
+  type: "ReLU6"
+  bottom: "conv2-1/dwise/scale"
+  top: "conv2-1/dwise/scale"
+}
 
 layer {
   name: "conv2-1/linear"
@@ -679,51 +679,12 @@ layer {
     bias_term: true
   }
 }
-# H-swish
-layer {
-  name: "conv4-2/dwise/shift"
-  type: "Power"
-  bottom: "conv4-2/dwise"
-  top: "conv4-2/dwise/shift"
-  power_param {
-    power: 1
-    scale: 1
-    shift: 3
-  }
-}
-layer {
-  name: "conv4-2/dwise/shift-relu"
-  type: "ReLU6"
-  bottom: "conv4-2/dwise/shift"
-  top: "conv4-2/dwise/shift"
-}
-layer {
-  name: "conv4-2/dwise/shift-div"
-  type: "Power"
-  bottom: "conv4-2/dwise/shift"
-  top: "conv4-2/dwise/shift-div"
-  power_param {
-    power: 1
-    scale: 0.1666666667
-    shift: 0
-  }
-}
-layer {
-  name: "conv4-2/dwise/hswish"
-  type: "Eltwise"
-  bottom: "conv4-2/dwise"
-  bottom: "conv4-2/dwise/shift-div"
-  top: "conv4-2/dwise/hswish"
-  eltwise_param {
-    operation: PROD
-  }
-}
 
 # se
 layer {
   name: "conv4-2/dwise-se-pool"
   type: "Pooling"
-  bottom: "conv4-2/dwise/hswish"
+  bottom: "conv4-2/dwise"
   top: "conv4-2/dwise-se-pool"
   pooling_param {
     pool: AVE
@@ -816,18 +777,57 @@ layer {
 layer {
   name: "conv4-2/dwise/scale"
   type: "Scale"
-  bottom: "conv4-2/dwise/hswish"
+  bottom: "conv4-2/dwise"
   bottom: "conv4-2/dwise-se-fc2/shift-div"
   top: "conv4-2/dwise/scale"
   scale_param{
     axis: 0
   }
 }
+# H-swish
+layer {
+  name: "conv4-2/dwise/shift"
+  type: "Power"
+  bottom: "conv4-2/dwise/scale"
+  top: "conv4-2/dwise/shift"
+  power_param {
+    power: 1
+    scale: 1
+    shift: 3
+  }
+}
+layer {
+  name: "conv4-2/dwise/shift-relu"
+  type: "ReLU6"
+  bottom: "conv4-2/dwise/shift"
+  top: "conv4-2/dwise/shift"
+}
+layer {
+  name: "conv4-2/dwise/shift-div"
+  type: "Power"
+  bottom: "conv4-2/dwise/shift"
+  top: "conv4-2/dwise/shift-div"
+  power_param {
+    power: 1
+    scale: 0.1666666667
+    shift: 0
+  }
+}
+layer {
+  name: "conv4-2/dwise/hswish"
+  type: "Eltwise"
+  bottom: "conv4-2/dwise/scale"
+  bottom: "conv4-2/dwise/shift-div"
+  top: "conv4-2/dwise/hswish"
+  eltwise_param {
+    operation: PROD
+  }
+}
 
 layer {
   name: "conv4-2/linear"
   type: "Convolution"
-  bottom: "conv4-2/dwise/scale"
+  bottom: "conv4-2/dwise/hswish"
   top: "conv4-2/linear"
   param {
     lr_mult: 1
@@ -974,51 +974,12 @@ layer {
     bias_term: true
   }
 }
-# H-swish
-layer {
-  name: "conv5-1/dwise/shift"
-  type: "Power"
-  bottom: "conv5-1/dwise"
-  top: "conv5-1/dwise/shift"
-  power_param {
-    power: 1
-    scale: 1
-    shift: 3
-  }
-}
-layer {
-  name: "conv5-1/dwise/shift-relu"
-  type: "ReLU6"
-  bottom: "conv5-1/dwise/shift"
-  top: "conv5-1/dwise/shift"
-}
-layer {
-  name: "conv5-1/dwise/shift-div"
-  type: "Power"
-  bottom: "conv5-1/dwise/shift"
-  top: "conv5-1/dwise/shift-div"
-  power_param {
-    power: 1
-    scale: 0.1666666667
-    shift: 0
-  }
-}
-layer {
-  name: "conv5-1/dwise/hswish"
-  type: "Eltwise"
-  bottom: "conv5-1/dwise"
-  bottom: "conv5-1/dwise/shift-div"
-  top: "conv5-1/dwise/hswish"
-  eltwise_param {
-    operation: PROD
-  }
-}
 
 # se
 layer {
   name: "conv5-1/dwise-se-pool"
   type: "Pooling"
-  bottom: "conv5-1/dwise/hswish"
+  bottom: "conv5-1/dwise"
   top: "conv5-1/dwise-se-pool"
   pooling_param {
     pool: AVE
@@ -1111,18 +1072,57 @@ layer {
 layer {
   name: "conv5-1/dwise/scale"
   type: "Scale"
-  bottom: "conv5-1/dwise/hswish"
+  bottom: "conv5-1/dwise"
   bottom: "conv5-1/dwise-se-fc2/shift-div"
   top: "conv5-1/dwise/scale"
   scale_param{
     axis: 0
   }
 }
+# H-swish
+layer {
+  name: "conv5-1/dwise/shift"
+  type: "Power"
+  bottom: "conv5-1/dwise/scale"
+  top: "conv5-1/dwise/shift"
+  power_param {
+    power: 1
+    scale: 1
+    shift: 3
+  }
+}
+layer {
+  name: "conv5-1/dwise/shift-relu"
+  type: "ReLU6"
+  bottom: "conv5-1/dwise/shift"
+  top: "conv5-1/dwise/shift"
+}
+layer {
+  name: "conv5-1/dwise/shift-div"
+  type: "Power"
+  bottom: "conv5-1/dwise/shift"
+  top: "conv5-1/dwise/shift-div"
+  power_param {
+    power: 1
+    scale: 0.1666666667
+    shift: 0
+  }
+}
+layer {
+  name: "conv5-1/dwise/hswish"
+  type: "Eltwise"
+  bottom: "conv5-1/dwise/scale"
+  bottom: "conv5-1/dwise/shift-div"
+  top: "conv5-1/dwise/hswish"
+  eltwise_param {
+    operation: PROD
+  }
+}
 
 layer {
   name: "conv5-1/linear"
   type: "Convolution"
-  bottom: "conv5-1/dwise/scale"
+  bottom: "conv5-1/dwise/hswish"
   top: "conv5-1/linear"
   param {
     lr_mult: 1
@@ -1277,51 +1277,12 @@ layer {
     bias_term: true
   }
 }
-# H-swish
-layer {
-  name: "conv5-2/dwise/shift"
-  type: "Power"
-  bottom: "conv5-2/dwise"
-  top: "conv5-2/dwise/shift"
-  power_param {
-    power: 1
-    scale: 1
-    shift: 3
-  }
-}
-layer {
-  name: "conv5-2/dwise/shift-relu"
-  type: "ReLU6"
-  bottom: "conv5-2/dwise/shift"
-  top: "conv5-2/dwise/shift"
-}
-layer {
-  name: "conv5-2/dwise/shift-div"
-  type: "Power"
-  bottom: "conv5-2/dwise/shift"
-  top: "conv5-2/dwise/shift-div"
-  power_param {
-    power: 1
-    scale: 0.1666666667
-    shift: 0
-  }
-}
-layer {
-  name: "conv5-2/dwise/hswish"
-  type: "Eltwise"
-  bottom: "conv5-2/dwise"
-  bottom: "conv5-2/dwise/shift-div"
-  top: "conv5-2/dwise/hswish"
-  eltwise_param {
-    operation: PROD
-  }
-}
 
 # se
 layer {
   name: "conv5-2/dwise-se-pool"
   type: "Pooling"
-  bottom: "conv5-2/dwise/hswish"
+  bottom: "conv5-2/dwise"
   top: "conv5-2/dwise-se-pool"
   pooling_param {
     pool: AVE
@@ -1414,43 +1375,82 @@ layer {
 layer {
   name: "conv5-2/dwise/scale"
   type: "Scale"
-  bottom: "conv5-2/dwise/hswish"
+  bottom: "conv5-2/dwise"
   bottom: "conv5-2/dwise-se-fc2/shift-div"
   top: "conv5-2/dwise/scale"
   scale_param{
     axis: 0
   }
 }
-
+# H-swish
 layer {
-  name: "conv5-2/linear"
-  type: "Convolution"
+  name: "conv5-2/dwise/shift"
+  type: "Power"
   bottom: "conv5-2/dwise/scale"
-  top: "conv5-2/linear"
-  param {
-    lr_mult: 1
-    decay_mult: 1
-  }
-  convolution_param {
-    num_output: 40
-    pad: 0
-    bias_term: false
-    kernel_size: 1
-    stride: 1
-    weight_filler {
-      type: "msra"
-    }
+  top: "conv5-2/dwise/shift"
+  power_param {
+    power: 1
+    scale: 1
+    shift: 3
   }
 }
 layer {
-  name: "conv5-2/linear-bn"
-  type: "BatchNorm"
-  bottom: "conv5-2/linear"
-  top: "conv5-2/linear"
+  name: "conv5-2/dwise/shift-relu"
+  type: "ReLU6"
+  bottom: "conv5-2/dwise/shift"
+  top: "conv5-2/dwise/shift"
 }
 layer {
-  name: "conv5-2/linear-bn-scale"
-  type: "Scale"
+  name: "conv5-2/dwise/shift-div"
+  type: "Power"
+  bottom: "conv5-2/dwise/shift"
+  top: "conv5-2/dwise/shift-div"
+  power_param {
+    power: 1
+    scale: 0.1666666667
+    shift: 0
+  }
+}
+layer {
+  name: "conv5-2/dwise/hswish"
+  type: "Eltwise"
+  bottom: "conv5-2/dwise/scale"
+  bottom: "conv5-2/dwise/shift-div"
+  top: "conv5-2/dwise/hswish"
+  eltwise_param {
+    operation: PROD
+  }
+}
+
+layer {
+  name: "conv5-2/linear"
+  type: "Convolution"
+  bottom: "conv5-2/dwise/hswish"
+  top: "conv5-2/linear"
+  param {
+    lr_mult: 1
+    decay_mult: 1
+  }
+  convolution_param {
+    num_output: 40
+    pad: 0
+    bias_term: false
+    kernel_size: 1
+    stride: 1
+    weight_filler {
+      type: "msra"
+    }
+  }
+}
+layer {
+  name: "conv5-2/linear-bn"
+  type: "BatchNorm"
+  bottom: "conv5-2/linear"
+  top: "conv5-2/linear"
+}
+layer {
+  name: "conv5-2/linear-bn-scale"
+  type: "Scale"
   bottom: "conv5-2/linear"
   top: "conv5-2/linear"
   scale_param {
@@ -1580,51 +1580,12 @@ layer {
     bias_term: true
   }
 }
-# H-swish
-layer {
-  name: "conv5-3/dwise/shift"
-  type: "Power"
-  bottom: "conv5-3/dwise"
-  top: "conv5-3/dwise/shift"
-  power_param {
-    power: 1
-    scale: 1
-    shift: 3
-  }
-}
-layer {
-  name: "conv5-3/dwise/shift-relu"
-  type: "ReLU6"
-  bottom: "conv5-3/dwise/shift"
-  top: "conv5-3/dwise/shift"
-}
-layer {
-  name: "conv5-3/dwise/shift-div"
-  type: "Power"
-  bottom: "conv5-3/dwise/shift"
-  top: "conv5-3/dwise/shift-div"
-  power_param {
-    power: 1
-    scale: 0.1666666667
-    shift: 0
-  }
-}
-layer {
-  name: "conv5-3/dwise/hswish"
-  type: "Eltwise"
-  bottom: "conv5-3/dwise"
-  bottom: "conv5-3/dwise/shift-div"
-  top: "conv5-3/dwise/hswish"
-  eltwise_param {
-    operation: PROD
-  }
-}
 
 # se
 layer {
   name: "conv5-3/dwise-se-pool"
   type: "Pooling"
-  bottom: "conv5-3/dwise/hswish"
+  bottom: "conv5-3/dwise"
   top: "conv5-3/dwise-se-pool"
   pooling_param {
     pool: AVE
@@ -1717,18 +1678,57 @@ layer {
 layer {
   name: "conv5-3/dwise/scale"
   type: "Scale"
-  bottom: "conv5-3/dwise/hswish"
+  bottom: "conv5-3/dwise"
   bottom: "conv5-3/dwise-se-fc2/shift-div"
   top: "conv5-3/dwise/scale"
   scale_param{
     axis: 0
   }
 }
+# H-swish
+layer {
+  name: "conv5-3/dwise/shift"
+  type: "Power"
+  bottom: "conv5-3/dwise/scale"
+  top: "conv5-3/dwise/shift"
+  power_param {
+    power: 1
+    scale: 1
+    shift: 3
+  }
+}
+layer {
+  name: "conv5-3/dwise/shift-relu"
+  type: "ReLU6"
+  bottom: "conv5-3/dwise/shift"
+  top: "conv5-3/dwise/shift"
+}
+layer {
+  name: "conv5-3/dwise/shift-div"
+  type: "Power"
+  bottom: "conv5-3/dwise/shift"
+  top: "conv5-3/dwise/shift-div"
+  power_param {
+    power: 1
+    scale: 0.1666666667
+    shift: 0
+  }
+}
+layer {
+  name: "conv5-3/dwise/hswish"
+  type: "Eltwise"
+  bottom: "conv5-3/dwise/scale"
+  bottom: "conv5-3/dwise/shift-div"
+  top: "conv5-3/dwise/hswish"
+  eltwise_param {
+    operation: PROD
+  }
+}
 
 layer {
   name: "conv5-3/linear"
   type: "Convolution"
-  bottom: "conv5-3/dwise/scale"
+  bottom: "conv5-3/dwise/hswish"
   top: "conv5-3/linear"
   param {
     lr_mult: 1
@@ -1875,51 +1875,12 @@ layer {
     bias_term: true
   }
 }
-# H-swish
-layer {
-  name: "conv5-4/dwise/shift"
-  type: "Power"
-  bottom: "conv5-4/dwise"
-  top: "conv5-4/dwise/shift"
-  power_param {
-    power: 1
-    scale: 1
-    shift: 3
-  }
-}
-layer {
-  name: "conv5-4/dwise/shift-relu"
-  type: "ReLU6"
-  bottom: "conv5-4/dwise/shift"
-  top: "conv5-4/dwise/shift"
-}
-layer {
-  name: "conv5-4/dwise/shift-div"
-  type: "Power"
-  bottom: "conv5-4/dwise/shift"
-  top: "conv5-4/dwise/shift-div"
-  power_param {
-    power: 1
-    scale: 0.1666666667
-    shift: 0
-  }
-}
-layer {
-  name: "conv5-4/dwise/hswish"
-  type: "Eltwise"
-  bottom: "conv5-4/dwise"
-  bottom: "conv5-4/dwise/shift-div"
-  top: "conv5-4/dwise/hswish"
-  eltwise_param {
-    operation: PROD
-  }
-}
 
 # se
 layer {
   name: "conv5-4/dwise-se-pool"
   type: "Pooling"
-  bottom: "conv5-4/dwise/hswish"
+  bottom: "conv5-4/dwise"
   top: "conv5-4/dwise-se-pool"
   pooling_param {
     pool: AVE
@@ -2012,18 +1973,57 @@ layer {
 layer {
   name: "conv5-4/dwise/scale"
   type: "Scale"
-  bottom: "conv5-4/dwise/hswish"
+  bottom: "conv5-4/dwise"
   bottom: "conv5-4/dwise-se-fc2/shift-div"
   top: "conv5-4/dwise/scale"
   scale_param{
     axis: 0
   }
 }
+# H-swish
+layer {
+  name: "conv5-4/dwise/shift"
+  type: "Power"
+  bottom: "conv5-4/dwise/scale"
+  top: "conv5-4/dwise/shift"
+  power_param {
+    power: 1
+    scale: 1
+    shift: 3
+  }
+}
+layer {
+  name: "conv5-4/dwise/shift-relu"
+  type: "ReLU6"
+  bottom: "conv5-4/dwise/shift"
+  top: "conv5-4/dwise/shift"
+}
+layer {
+  name: "conv5-4/dwise/shift-div"
+  type: "Power"
+  bottom: "conv5-4/dwise/shift"
+  top: "conv5-4/dwise/shift-div"
+  power_param {
+    power: 1
+    scale: 0.1666666667
+    shift: 0
+  }
+}
+layer {
+  name: "conv5-4/dwise/hswish"
+  type: "Eltwise"
+  bottom: "conv5-4/dwise/scale"
+  bottom: "conv5-4/dwise/shift-div"
+  top: "conv5-4/dwise/hswish"
+  eltwise_param {
+    operation: PROD
+  }
+}
 
 layer {
   name: "conv5-4/linear"
   type: "Convolution"
-  bottom: "conv5-4/dwise/scale"
+  bottom: "conv5-4/dwise/hswish"
   top: "conv5-4/linear"
   param {
     lr_mult: 1
@@ -2178,51 +2178,12 @@ layer {
     bias_term: true
   }
 }
-# H-swish
-layer {
-  name: "conv5-5/dwise/shift"
-  type: "Power"
-  bottom: "conv5-5/dwise"
-  top: "conv5-5/dwise/shift"
-  power_param {
-    power: 1
-    scale: 1
-    shift: 3
-  }
-}
-layer {
-  name: "conv5-5/dwise/shift-relu"
-  type: "ReLU6"
-  bottom: "conv5-5/dwise/shift"
-  top: "conv5-5/dwise/shift"
-}
-layer {
-  name: "conv5-5/dwise/shift-div"
-  type: "Power"
-  bottom: "conv5-5/dwise/shift"
-  top: "conv5-5/dwise/shift-div"
-  power_param {
-    power: 1
-    scale: 0.1666666667
-    shift: 0
-  }
-}
-layer {
-  name: "conv5-5/dwise/hswish"
-  type: "Eltwise"
-  bottom: "conv5-5/dwise"
-  bottom: "conv5-5/dwise/shift-div"
-  top: "conv5-5/dwise/hswish"
-  eltwise_param {
-    operation: PROD
-  }
-}
 
 # se
 layer {
   name: "conv5-5/dwise-se-pool"
   type: "Pooling"
-  bottom: "conv5-5/dwise/hswish"
+  bottom: "conv5-5/dwise"
   top: "conv5-5/dwise-se-pool"
   pooling_param {
     pool: AVE
@@ -2315,18 +2276,57 @@ layer {
 layer {
   name: "conv5-5/dwise/scale"
   type: "Scale"
-  bottom: "conv5-5/dwise/hswish"
+  bottom: "conv5-5/dwise"
   bottom: "conv5-5/dwise-se-fc2/shift-div"
   top: "conv5-5/dwise/scale"
   scale_param{
     axis: 0
   }
 }
+# H-swish
+layer {
+  name: "conv5-5/dwise/shift"
+  type: "Power"
+  bottom: "conv5-5/dwise/scale"
+  top: "conv5-5/dwise/shift"
+  power_param {
+    power: 1
+    scale: 1
+    shift: 3
+  }
+}
+layer {
+  name: "conv5-5/dwise/shift-relu"
+  type: "ReLU6"
+  bottom: "conv5-5/dwise/shift"
+  top: "conv5-5/dwise/shift"
+}
+layer {
+  name: "conv5-5/dwise/shift-div"
+  type: "Power"
+  bottom: "conv5-5/dwise/shift"
+  top: "conv5-5/dwise/shift-div"
+  power_param {
+    power: 1
+    scale: 0.1666666667
+    shift: 0
+  }
+}
+layer {
+  name: "conv5-5/dwise/hswish"
+  type: "Eltwise"
+  bottom: "conv5-5/dwise/scale"
+  bottom: "conv5-5/dwise/shift-div"
+  top: "conv5-5/dwise/hswish"
+  eltwise_param {
+    operation: PROD
+  }
+}
 
 layer {
   name: "conv5-5/linear"
   type: "Convolution"
-  bottom: "conv5-5/dwise/scale"
+  bottom: "conv5-5/dwise/hswish"
   top: "conv5-5/linear"
   param {
     lr_mult: 1
@@ -2473,51 +2473,12 @@ layer {
     bias_term: true
   }
 }
-# H-swish
-layer {
-  name: "conv6-1/dwise/shift"
-  type: "Power"
-  bottom: "conv6-1/dwise"
-  top: "conv6-1/dwise/shift"
-  power_param {
-    power: 1
-    scale: 1
-    shift: 3
-  }
-}
-layer {
-  name: "conv6-1/dwise/shift-relu"
-  type: "ReLU6"
-  bottom: "conv6-1/dwise/shift"
-  top: "conv6-1/dwise/shift"
-}
-layer {
-  name: "conv6-1/dwise/shift-div"
-  type: "Power"
-  bottom: "conv6-1/dwise/shift"
-  top: "conv6-1/dwise/shift-div"
-  power_param {
-    power: 1
-    scale: 0.1666666667
-    shift: 0
-  }
-}
-layer {
-  name: "conv6-1/dwise/hswish"
-  type: "Eltwise"
-  bottom: "conv6-1/dwise"
-  bottom: "conv6-1/dwise/shift-div"
-  top: "conv6-1/dwise/hswish"
-  eltwise_param {
-    operation: PROD
-  }
-}
 
 # se
 layer {
   name: "conv6-1/dwise-se-pool"
   type: "Pooling"
-  bottom: "conv6-1/dwise/hswish"
+  bottom: "conv6-1/dwise"
   top: "conv6-1/dwise-se-pool"
   pooling_param {
     pool: AVE
@@ -2610,18 +2571,57 @@ layer {
 layer {
   name: "conv6-1/dwise/scale"
   type: "Scale"
-  bottom: "conv6-1/dwise/hswish"
+  bottom: "conv6-1/dwise"
   bottom: "conv6-1/dwise-se-fc2/shift-div"
   top: "conv6-1/dwise/scale"
   scale_param{
     axis: 0
   }
 }
+# H-swish
+layer {
+  name: "conv6-1/dwise/shift"
+  type: "Power"
+  bottom: "conv6-1/dwise/scale"
+  top: "conv6-1/dwise/shift"
+  power_param {
+    power: 1
+    scale: 1
+    shift: 3
+  }
+}
+layer {
+  name: "conv6-1/dwise/shift-relu"
+  type: "ReLU6"
+  bottom: "conv6-1/dwise/shift"
+  top: "conv6-1/dwise/shift"
+}
+layer {
+  name: "conv6-1/dwise/shift-div"
+  type: "Power"
+  bottom: "conv6-1/dwise/shift"
+  top: "conv6-1/dwise/shift-div"
+  power_param {
+    power: 1
+    scale: 0.1666666667
+    shift: 0
+  }
+}
+layer {
+  name: "conv6-1/dwise/hswish"
+  type: "Eltwise"
+  bottom: "conv6-1/dwise/scale"
+  bottom: "conv6-1/dwise/shift-div"
+  top: "conv6-1/dwise/hswish"
+  eltwise_param {
+    operation: PROD
+  }
+}
 
 layer {
   name: "conv6-1/linear"
   type: "Convolution"
-  bottom: "conv6-1/dwise/scale"
+  bottom: "conv6-1/dwise/hswish"
   top: "conv6-1/linear"
   param {
     lr_mult: 1
@@ -2776,51 +2776,12 @@ layer {
     bias_term: true
   }
 }
-# H-swish
-layer {
-  name: "conv6-2/dwise/shift"
-  type: "Power"
-  bottom: "conv6-2/dwise"
-  top: "conv6-2/dwise/shift"
-  power_param {
-    power: 1
-    scale: 1
-    shift: 3
-  }
-}
-layer {
-  name: "conv6-2/dwise/shift-relu"
-  type: "ReLU6"
-  bottom: "conv6-2/dwise/shift"
-  top: "conv6-2/dwise/shift"
-}
-layer {
-  name: "conv6-2/dwise/shift-div"
-  type: "Power"
-  bottom: "conv6-2/dwise/shift"
-  top: "conv6-2/dwise/shift-div"
-  power_param {
-    power: 1
-    scale: 0.1666666667
-    shift: 0
-  }
-}
-layer {
-  name: "conv6-2/dwise/hswish"
-  type: "Eltwise"
-  bottom: "conv6-2/dwise"
-  bottom: "conv6-2/dwise/shift-div"
-  top: "conv6-2/dwise/hswish"
-  eltwise_param {
-    operation: PROD
-  }
-}
 
 # se
 layer {
   name: "conv6-2/dwise-se-pool"
   type: "Pooling"
-  bottom: "conv6-2/dwise/hswish"
+  bottom: "conv6-2/dwise"
   top: "conv6-2/dwise-se-pool"
   pooling_param {
     pool: AVE
@@ -2913,18 +2874,57 @@ layer {
 layer {
   name: "conv6-2/dwise/scale"
   type: "Scale"
-  bottom: "conv6-2/dwise/hswish"
+  bottom: "conv6-2/dwise"
   bottom: "conv6-2/dwise-se-fc2/shift-div"
   top: "conv6-2/dwise/scale"
   scale_param{
     axis: 0
   }
 }
+# H-swish
+layer {
+  name: "conv6-2/dwise/shift"
+  type: "Power"
+  bottom: "conv6-2/dwise/scale"
+  top: "conv6-2/dwise/shift"
+  power_param {
+    power: 1
+    scale: 1
+    shift: 3
+  }
+}
+layer {
+  name: "conv6-2/dwise/shift-relu"
+  type: "ReLU6"
+  bottom: "conv6-2/dwise/shift"
+  top: "conv6-2/dwise/shift"
+}
+layer {
+  name: "conv6-2/dwise/shift-div"
+  type: "Power"
+  bottom: "conv6-2/dwise/shift"
+  top: "conv6-2/dwise/shift-div"
+  power_param {
+    power: 1
+    scale: 0.1666666667
+    shift: 0
+  }
+}
+layer {
+  name: "conv6-2/dwise/hswish"
+  type: "Eltwise"
+  bottom: "conv6-2/dwise/scale"
+  bottom: "conv6-2/dwise/shift-div"
+  top: "conv6-2/dwise/hswish"
+  eltwise_param {
+    operation: PROD
+  }
+}
 
 layer {
   name: "conv6-2/linear"
   type: "Convolution"
-  bottom: "conv6-2/dwise/scale"
+  bottom: "conv6-2/dwise/hswish"
   top: "conv6-2/linear"
   param {
     lr_mult: 1
@@ -3001,51 +3001,12 @@ layer {
     bias_term: true
   }
 }
-# H-swish
-layer {
-  name: "conv6-3/shift"
-  type: "Power"
-  bottom: "conv6-3"
-  top: "conv6-3/shift"
-  power_param {
-    power: 1
-    scale: 1
-    shift: 3
-  }
-}
-layer {
-  name: "conv6-3/shift-relu"
-  type: "ReLU6"
-  bottom: "conv6-3/shift"
-  top: "conv6-3/shift"
-}
-layer {
-  name: "conv6-3/shift-div"
-  type: "Power"
-  bottom: "conv6-3/shift"
-  top: "conv6-3/shift-div"
-  power_param {
-    power: 1
-    scale: 0.1666666667
-    shift: 0
-  }
-}
-layer {
-  name: "conv6-3/hswish"
-  type: "Eltwise"
-  bottom: "conv6-3"
-  bottom: "conv6-3/shift-div"
-  top: "conv6-3/hswish"
-  eltwise_param {
-    operation: PROD
-  }
-}
 
 # se
 layer {
   name: "conv6-3-se-pool"
   type: "Pooling"
-  bottom: "conv6-3/hswish"
+  bottom: "conv6-3"
   top: "conv6-3-se-pool"
   pooling_param {
     pool: AVE
@@ -3138,19 +3099,58 @@ layer {
 layer {
   name: "conv6-3/scale"
   type: "Scale"
-  bottom: "conv6-3/hswish"
+  bottom: "conv6-3"
   bottom: "conv6-3-se-fc2/shift-div"
   top: "conv6-3/scale"
   scale_param{
     axis: 0
   }
 }
+# H-swish
+layer {
+  name: "conv6-3/shift"
+  type: "Power"
+  bottom: "conv6-3/scale"
+  top: "conv6-3/shift"
+  power_param {
+    power: 1
+    scale: 1
+    shift: 3
+  }
+}
+layer {
+  name: "conv6-3/shift-relu"
+  type: "ReLU6"
+  bottom: "conv6-3/shift"
+  top: "conv6-3/shift"
+}
+layer {
+  name: "conv6-3/shift-div"
+  type: "Power"
+  bottom: "conv6-3/shift"
+  top: "conv6-3/shift-div"
+  power_param {
+    power: 1
+    scale: 0.1666666667
+    shift: 0
+  }
+}
+layer {
+  name: "conv6-3/hswish"
+  type: "Eltwise"
+  bottom: "conv6-3/scale"
+  bottom: "conv6-3/shift-div"
+  top: "conv6-3/hswish"
+  eltwise_param {
+    operation: PROD
+  }
+}
 
 #################### stage 7-1 ####################
 layer {
   name: "pool7-1"
   type: "Pooling"
-  bottom: "conv6-3/scale"
+  bottom: "conv6-3/hswish"
   top: "pool7-1"
   pooling_param {
     pool: AVE