Use parameter default value to control default behavior of init_cfg in (#319)

`LinearClsHead` and `MultiLabelLinearClsHead`

And remove the verbose `_init_layers` method of `LinearClsHead` and
`MultiLabelLinearClsHead`.

mmcls/models/heads/linear_head.py

@@ -14,20 +14,16 @@ class LinearClsHead(ClsHead):
         num_classes (int): Number of categories excluding the background
             category.
         in_channels (int): Number of channels in the input feature map.
+        init_cfg (dict | optional): The extra init config of layers.
+            Defaults to use dict(type='Normal', layer='Linear', std=0.01).
     """
 
     def __init__(self,
                  num_classes,
                  in_channels,
-                 init_cfg=None,
+                 init_cfg=dict(type='Normal', layer='Linear', std=0.01),
                  *args,
                  **kwargs):
-        init_cfg = init_cfg or dict(
-            type='Normal',
-            mean=0.,
-            std=0.01,
-            bias=0.,
-            override=dict(name='fc'))
         super(LinearClsHead, self).__init__(init_cfg=init_cfg, *args, **kwargs)
 
         self.in_channels = in_channels
@@ -37,9 +33,6 @@ class LinearClsHead(ClsHead):
             raise ValueError(
                 f'num_classes={num_classes} must be a positive integer')
 
-        self._init_layers()
-
-    def _init_layers(self):
         self.fc = nn.Linear(self.in_channels, self.num_classes)
 
     def simple_test(self, img):

mmcls/models/heads/multi_label_linear_head.py

@@ -14,6 +14,8 @@ class MultiLabelLinearClsHead(MultiLabelClsHead):
         num_classes (int): Number of categories.
         in_channels (int): Number of channels in the input feature map.
         loss (dict): Config of classification loss.
+        init_cfg (dict | optional): The extra init config of layers.
+            Defaults to use dict(type='Normal', layer='Linear', std=0.01).
     """
 
     def __init__(self,
@@ -24,12 +26,7 @@ class MultiLabelLinearClsHead(MultiLabelClsHead):
                      use_sigmoid=True,
                      reduction='mean',
                      loss_weight=1.0),
-                 init_cfg=dict(
-                     type='Normal',
-                     mean=0.,
-                     std=0.01,
-                     bias=0.,
-                     override=dict(name='fc'))):
+                 init_cfg=dict(type='Normal', layer='Linear', std=0.01)):
         super(MultiLabelLinearClsHead, self).__init__(
             loss=loss, init_cfg=init_cfg)
 
@@ -39,9 +36,7 @@ class MultiLabelLinearClsHead(MultiLabelClsHead):
 
         self.in_channels = in_channels
         self.num_classes = num_classes
-        self._init_layers()
 
-    def _init_layers(self):
         self.fc = nn.Linear(self.in_channels, self.num_classes)
 
     def forward_train(self, x, gt_label):