Fixed Typo in modeling_bart.py (#12035)

* Fixed Typo in modeling_bart.py - Issue #11895

* Fixed Typo in modeling_bart.py

src/transformers/models/bart/modeling_bart.py

@@ -370,10 +370,10 @@ class BartDecoderLayer(nn.Module):
     ):
         """
         Args:
-            hidden_states (:obj:`torch.FloatTensor`): input to the layer of shape `(seq_len, batch, embed_dim)`
+            hidden_states (:obj:`torch.FloatTensor`): input to the layer of shape `(batch, seq_len, embed_dim)`
             attention_mask (:obj:`torch.FloatTensor`): attention mask of size
                 `(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
-            encoder_hidden_states (:obj:`torch.FloatTensor`): cross attention input to the layer of shape `(seq_len, batch, embed_dim)`
+            encoder_hidden_states (:obj:`torch.FloatTensor`): cross attention input to the layer of shape `(batch, seq_len, embed_dim)`
             encoder_attention_mask (:obj:`torch.FloatTensor`): encoder attention mask of size
                 `(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
             layer_head_mask (:obj:`torch.FloatTensor`): mask for attention heads in a given layer of size

src/transformers/models/blenderbot_small/modeling_blenderbot_small.py

@@ -371,10 +371,10 @@ class BlenderbotSmallDecoderLayer(nn.Module):
     ):
         """
         Args:
-            hidden_states (:obj:`torch.FloatTensor`): input to the layer of shape `(seq_len, batch, embed_dim)`
+            hidden_states (:obj:`torch.FloatTensor`): input to the layer of shape `(batch, seq_len, embed_dim)`
             attention_mask (:obj:`torch.FloatTensor`): attention mask of size
                 `(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
-            encoder_hidden_states (:obj:`torch.FloatTensor`): cross attention input to the layer of shape `(seq_len, batch, embed_dim)`
+            encoder_hidden_states (:obj:`torch.FloatTensor`): cross attention input to the layer of shape `(batch, seq_len, embed_dim)`
             encoder_attention_mask (:obj:`torch.FloatTensor`): encoder attention mask of size
                 `(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
             layer_head_mask (:obj:`torch.FloatTensor`): mask for attention heads in a given layer of size

src/transformers/models/marian/modeling_marian.py

@@ -388,10 +388,10 @@ class MarianDecoderLayer(nn.Module):
     ):
         """
         Args:
-            hidden_states (:obj:`torch.FloatTensor`): input to the layer of shape `(seq_len, batch, embed_dim)`
+            hidden_states (:obj:`torch.FloatTensor`): input to the layer of shape `(batch, seq_len, embed_dim)`
             attention_mask (:obj:`torch.FloatTensor`): attention mask of size
                 `(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
-            encoder_hidden_states (:obj:`torch.FloatTensor`): cross attention input to the layer of shape `(seq_len, batch, embed_dim)`
+            encoder_hidden_states (:obj:`torch.FloatTensor`): cross attention input to the layer of shape `(batch, seq_len, embed_dim)`
             encoder_attention_mask (:obj:`torch.FloatTensor`): encoder attention mask of size
                 `(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
             layer_head_mask (:obj:`torch.FloatTensor`): mask for attention heads in a given layer of size