modify the prelu test and regenerate the proto

python/paddle/trainer_config_helpers/tests/configs/protostr/test_prelu_layer.protostr

@@ -4,6 +4,8 @@ layers {
   type: "data"
   size: 300
   active_type: ""
+  height: 10
+  width: 10
 }
 layers {
   name: "__prelu_layer_0__"
@@ -15,6 +17,9 @@ layers {
     input_parameter_name: "___prelu_layer_0__.w0"
   }
   partial_sum: 1
+  height: 10
+  width: 10
+  depth: 1
 }
 layers {
   name: "__prelu_layer_1__"
@@ -26,6 +31,9 @@ layers {
     input_parameter_name: "___prelu_layer_1__.w0"
   }
   partial_sum: 1
+  height: 10
+  width: 10
+  depth: 1
 }
 layers {
   name: "__prelu_layer_2__"
@@ -37,41 +45,100 @@ layers {
     input_parameter_name: "___prelu_layer_2__.w0"
   }
   partial_sum: 5
+  height: 10
+  width: 10
+  depth: 1
+}
+layers {
+  name: "__prelu_layer_3__"
+  type: "prelu"
+  size: 300
+  active_type: ""
+  inputs {
+    input_layer_name: "input"
+    input_parameter_name: "___prelu_layer_3__.w0"
+  }
+  partial_sum: 300
+  height: 10
+  width: 10
+  depth: 1
+}
+layers {
+  name: "__prelu_layer_4__"
+  type: "prelu"
+  size: 300
+  active_type: ""
+  inputs {
+    input_layer_name: "input"
+    input_parameter_name: "___prelu_layer_4__.w0"
+  }
+  partial_sum: 100
+  height: 10
+  width: 10
+  depth: 1
 }
 parameters {
   name: "___prelu_layer_0__.w0"
   size: 300
-  initial_mean: 0.0
-  initial_std: 0.057735026919
+  initial_mean: 0.25
+  initial_std: 0.0
+  dims: 1
+  dims: 300
   initial_strategy: 0
-  initial_smart: true
+  initial_smart: false
 }
 parameters {
   name: "___prelu_layer_1__.w0"
   size: 300
-  initial_mean: 0.0
-  initial_std: 0.057735026919
+  initial_mean: 0.25
+  initial_std: 0.0
+  dims: 1
+  dims: 300
   initial_strategy: 0
-  initial_smart: true
+  initial_smart: false
 }
 parameters {
   name: "___prelu_layer_2__.w0"
   size: 60
-  initial_mean: 0.0
-  initial_std: 0.129099444874
+  initial_mean: 0.25
+  initial_std: 0.0
+  dims: 1
+  dims: 60
+  initial_strategy: 0
+  initial_smart: false
+}
+parameters {
+  name: "___prelu_layer_3__.w0"
+  size: 1
+  initial_mean: 0.25
+  initial_std: 0.0
+  dims: 1
+  dims: 1
+  initial_strategy: 0
+  initial_smart: false
+}
+parameters {
+  name: "___prelu_layer_4__.w0"
+  size: 3
+  initial_mean: 0.25
+  initial_std: 0.0
+  dims: 1
+  dims: 3
   initial_strategy: 0
-  initial_smart: true
+  initial_smart: false
 }
 input_layer_names: "input"
-output_layer_names: "__prelu_layer_2__"
+output_layer_names: "__prelu_layer_4__"
 sub_models {
   name: "root"
   layer_names: "input"
   layer_names: "__prelu_layer_0__"
   layer_names: "__prelu_layer_1__"
   layer_names: "__prelu_layer_2__"
+  layer_names: "__prelu_layer_3__"
+  layer_names: "__prelu_layer_4__"
   input_layer_names: "input"
-  output_layer_names: "__prelu_layer_2__"
+  output_layer_names: "__prelu_layer_4__"
   is_recurrent_layer_group: false
 }
 

python/paddle/trainer_config_helpers/tests/configs/test_prelu_layer.py

 from paddle.trainer_config_helpers import *
 
-data = data_layer(name='input', size=300)
-prelu = prelu_layer(input=data)
-prelu = prelu_layer(input=data, partial_sum=1)
-prelu = prelu_layer(input=data, partial_sum=5)
+data = data_layer(name='input', size=300, height=10, width=10)
+prelu = prelu_layer(input=data, num_channels=3)
+prelu = prelu_layer(input=data, partial_sum=1, num_channels=3)
+prelu = prelu_layer(input=data, partial_sum=5, num_channels=3)
+prelu = prelu_layer(input=data, channel_shared=True, num_channels=3)
+prelu = prelu_layer(input=data, channel_shared=False, num_channels=3)
 
 outputs(prelu)