micro_block -> layer_type

ppcls/arch/backbone/legendary_models/resnet.py

@@ -164,18 +164,18 @@ class BottleneckBlock(TheseusLayer):
                  stride,
                  shortcut=True,
                  if_first=False,
-                 micro_block=ConvBNLayer,
+                 layer=ConvBNLayer,
                  lr_mult=1.0,
                  data_format="NCHW"):
         super().__init__()
-        self.conv0 = micro_block(
+        self.conv0 = layer(
             num_channels=num_channels,
             num_filters=num_filters,
             filter_size=1,
             act="relu",
             lr_mult=lr_mult,
             data_format=data_format)
-        self.conv1 = micro_block(
+        self.conv1 = layer(
             num_channels=num_filters,
             num_filters=num_filters,
             filter_size=3,
@@ -183,7 +183,7 @@ class BottleneckBlock(TheseusLayer):
             act="relu",
             lr_mult=lr_mult,
             data_format=data_format)
-        self.conv2 = micro_block(
+        self.conv2 = layer(
             num_channels=num_filters,
             num_filters=num_filters * 4,
             filter_size=1,
@@ -226,13 +226,13 @@ class BasicBlock(TheseusLayer):
                  stride,
                  shortcut=True,
                  if_first=False,
-                 micro_block=ConvBNLayer,
+                 layer=ConvBNLayer,
                  lr_mult=1.0,
                  data_format="NCHW"):
         super().__init__()
 
         self.stride = stride
-        self.conv0 = micro_block(
+        self.conv0 = layer(
             num_channels=num_channels,
             num_filters=num_filters,
             filter_size=3,
@@ -240,7 +240,7 @@ class BasicBlock(TheseusLayer):
             act="relu",
             lr_mult=lr_mult,
             data_format=data_format)
-        self.conv1 = micro_block(
+        self.conv1 = layer(
             num_channels=num_filters,
             num_filters=num_filters,
             filter_size=3,
@@ -296,7 +296,7 @@ class ResNet(TheseusLayer):
                  input_image_channel=3,
                  return_patterns=None,
                  return_stages=None,
-                 micro_block="ConvBNLayer",
+                 layer_type="ConvBNLayer",
                  use_first_short_conv=True,
                  **kargs):
         super().__init__()
@@ -312,10 +312,10 @@ class ResNet(TheseusLayer):
         self.num_channels = self.cfg["num_channels"]
         self.channels_mult = 1 if self.num_channels[-1] == 256 else 4
 
-        if micro_block == "ConvBNLayer":
-            micro_block = ConvBNLayer
-        elif micro_block == "DiverseBranchBlock":
-            micro_block = DiverseBranchBlock
+        if layer_type == "ConvBNLayer":
+            layer = ConvBNLayer
+        elif layer_type == "DiverseBranchBlock":
+            layer = DiverseBranchBlock
         else:
             raise Exception()
 
@@ -377,7 +377,7 @@ class ResNet(TheseusLayer):
                     if i == 0 and block_idx != 0 else 1,
                     shortcut=shortcut,
                     if_first=block_idx == i == 0 if version == "vd" else True,
-                    micro_block=micro_block,
+                    layer=layer,
                     lr_mult=self.lr_mult_list[block_idx + 1],
                     data_format=data_format))
                 shortcut = True

ppcls/configs/ImageNet/ResNet/ResNet18_dbb.yaml

@@ -19,7 +19,7 @@ Global:
 Arch:
   name: ResNet18
   class_num: 1000
-  micro_block: DiverseBranchBlock
+  layer_type: DiverseBranchBlock
   use_first_short_conv: False