Fix initilizations (#1)

* fix: code structure in few cases.

* fix: code structure to align tf models.

* fix: layer naming, bn layer still remains.

* chore: change default epsilon and momentum in bn.

src/transformers/modeling_tf_outputs.py

@@ -110,8 +110,8 @@ class TFBaseModelOutputWithPoolingAndNoAttention(ModelOutput):
         pooler_output (`tf.Tensor` of shape `(batch_size, hidden_size)`):
             Last layer hidden-state after a pooling operation on the spatial dimensions.
         hidden_states (`tuple(tf.Tensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
-            Tuple of `tf.Tensor` (one for the output of the embeddings, if the model has an embedding layer, +
-            one for the output of each layer) of shape `(batch_size, num_channels, height, width)`.
+            Tuple of `tf.Tensor` (one for the output of the embeddings, if the model has an embedding layer, + one for
+            the output of each layer) of shape `(batch_size, num_channels, height, width)`.
 
             Hidden-states of the model at the output of each layer plus the optional initial embedding outputs.
     """
@@ -830,6 +830,7 @@ class TFSequenceClassifierOutputWithPast(ModelOutput):
     hidden_states: Optional[Tuple[tf.Tensor]] = None
     attentions: Optional[Tuple[tf.Tensor]] = None
 
+
 @dataclass
 class TFImageClassifierOutputWithNoAttention(ModelOutput):
     """
@@ -841,9 +842,9 @@ class TFImageClassifierOutputWithNoAttention(ModelOutput):
         logits (`tf.Tensor` of shape `(batch_size, config.num_labels)`):
             Classification (or regression if config.num_labels==1) scores (before SoftMax).
         hidden_states (`tuple(tf.Tensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
-            Tuple of `tf.Tensor` (one for the output of the embeddings, if the model has an embedding layer, +
-            one for the output of each stage) of shape `(batch_size, num_channels, height, width)`. Hidden-states (also
-            called feature maps) of the model at the output of each stage.
+            Tuple of `tf.Tensor` (one for the output of the embeddings, if the model has an embedding layer, + one for
+            the output of each stage) of shape `(batch_size, num_channels, height, width)`. Hidden-states (also called
+            feature maps) of the model at the output of each stage.
     """
 
     loss: Optional[tf.Tensor] = None

src/transformers/models/regnet/configuration_regnet.py

@@ -36,6 +36,8 @@ class RegNetConfig(PretrainedConfig):
     documentation from [`PretrainedConfig`] for more information.
 
     Args:
+        image_size (`int`, *optional*, defaults to 224):
+            Size of the input images.
         num_channels (`int`, *optional*, defaults to 3):
             The number of input channels.
         embedding_size (`int`, *optional*, defaults to 64):
@@ -71,6 +73,7 @@ class RegNetConfig(PretrainedConfig):
 
     def __init__(
         self,
+        image_size=224,
         num_channels=3,
         embedding_size=32,
         hidden_sizes=[128, 192, 512, 1088],
@@ -83,6 +86,7 @@ def __init__(
         super().__init__(**kwargs)
         if layer_type not in self.layer_types:
             raise ValueError(f"layer_type={layer_type} is not one of {','.join(self.layer_types)}")
+        self.image_size = image_size
         self.num_channels = num_channels
         self.embedding_size = embedding_size
         self.hidden_sizes = hidden_sizes