!6329 Adjust GroupNorm interface

Merge pull request !6329 from liangchenghui/fix_groupnorm

mindspore/nn/layer/normalization.py

@@ -572,10 +572,10 @@ class GroupNorm(Cell):
         affine (bool): A bool value, this layer will have learnable affine parameters when set to true. Default: True.
         gamma_init (Union[Tensor, str, Initializer, numbers.Number]): Initializer for the gamma weight.
             The values of str refer to the function `initializer` including 'zeros', 'ones', 'xavier_uniform',
-            'he_uniform', etc. Default: 'ones'.
+            'he_uniform', etc. Default: 'ones'. If gamma_init is a Tensor, the shape must be [num_channels].
         beta_init (Union[Tensor, str, Initializer, numbers.Number]): Initializer for the beta weight.
             The values of str refer to the function `initializer` including 'zeros', 'ones', 'xavier_uniform',
-            'he_uniform', etc. Default: 'zeros'.
+            'he_uniform', etc. Default: 'zeros'. If beta_init is a Tensor, the shape must be [num_channels].
 
     Inputs:
         - **input_x** (Tensor) - The input feature with shape [N, C, H, W].
@@ -607,8 +607,8 @@ class GroupNorm(Cell):
         self.eps = check_typename('eps', eps, (float,))
         self.affine = check_bool(affine)
 
-        gamma = initializer(gamma_init, [num_channels, 1, 1])
-        beta = initializer(beta_init, [num_channels, 1, 1])
+        gamma = initializer(gamma_init, num_channels)
+        beta = initializer(beta_init, num_channels)
         if self.affine:
             self.gamma = Parameter(gamma, name='gamma')
             self.beta = Parameter(beta, name='beta')
@@ -632,7 +632,7 @@ class GroupNorm(Cell):
         std = self.sqrt(var + self.eps)
         x = (x - mean) / std
         x = self.reshape(x, (batch, channel, height, width))
-        output = x * self.gamma + self.beta
+        output = x * self.reshape(self.gamma, (-1, 1, 1)) + self.reshape(self.beta, (-1, 1, 1))
         return output
 
     def construct(self, x):