[S2T, Whisper] Add copied from statements (#20787)

* [S2T, Whisper] Add copied from statements

* rebase and fix-copies

src/transformers/models/speech_to_text/modeling_speech_to_text.py

@@ -354,6 +354,7 @@ class Speech2TextAttention(nn.Module):
         return attn_output, attn_weights_reshaped, past_key_value
 
 
+# Copied from transformers.models.mbart.modeling_mbart.MBartEncoderLayer with MBart->Speech2Text
 class Speech2TextEncoderLayer(nn.Module):
     def __init__(self, config: Speech2TextConfig):
         super().__init__()
@@ -377,14 +378,14 @@ class Speech2TextEncoderLayer(nn.Module):
         attention_mask: torch.Tensor,
         layer_head_mask: torch.Tensor,
         output_attentions: bool = False,
-    ):
+    ) -> torch.Tensor:
         """
         Args:
             hidden_states (`torch.FloatTensor`): input to the layer of shape `(seq_len, batch, embed_dim)`
             attention_mask (`torch.FloatTensor`): attention mask of size
                 `(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
             layer_head_mask (`torch.FloatTensor`): mask for attention heads in a given layer of size
-                `(config.encoder_attention_heads,)`.
+                `(encoder_attention_heads,)`.
             output_attentions (`bool`, *optional*):
                 Whether or not to return the attentions tensors of all attention layers. See `attentions` under
                 returned tensors for more detail.
@@ -422,6 +423,7 @@ class Speech2TextEncoderLayer(nn.Module):
         return outputs
 
 
+# Copied from transformers.models.mbart.modeling_mbart.MBartDecoderLayer with MBart->Speech2Text
 class Speech2TextDecoderLayer(nn.Module):
     def __init__(self, config: Speech2TextConfig):
         super().__init__()
@@ -460,7 +462,7 @@ class Speech2TextDecoderLayer(nn.Module):
         past_key_value: Optional[Tuple[torch.Tensor]] = None,
         output_attentions: Optional[bool] = False,
         use_cache: Optional[bool] = True,
-    ):
+    ) -> torch.Tensor:
         """
         Args:
             hidden_states (`torch.FloatTensor`): input to the layer of shape `(seq_len, batch, embed_dim)`
@@ -473,7 +475,7 @@ class Speech2TextDecoderLayer(nn.Module):
             layer_head_mask (`torch.FloatTensor`): mask for attention heads in a given layer of size
                 `(encoder_attention_heads,)`.
             cross_attn_layer_head_mask (`torch.FloatTensor`): mask for cross-attention heads in a given layer of
-                size *(decoder_attention_heads,)*.
+                size `(decoder_attention_heads,)`.
             past_key_value (`Tuple(torch.FloatTensor)`): cached past key and value projection states
             output_attentions (`bool`, *optional*):
                 Whether or not to return the attentions tensors of all attention layers. See `attentions` under

src/transformers/models/whisper/modeling_whisper.py

@@ -261,7 +261,7 @@ class WhisperAttention(nn.Module):
         return attn_output, attn_weights_reshaped, past_key_value
 
 
-# Copied from transformers.models.speech_to_text.modeling_speech_to_text.Speech2TextEncoderLayer with Speech2Text->Whisper
+# Copied from transformers.models.mbart.modeling_mbart.MBartEncoderLayer with MBart->Whisper
 class WhisperEncoderLayer(nn.Module):
     def __init__(self, config: WhisperConfig):
         super().__init__()
@@ -285,14 +285,14 @@ class WhisperEncoderLayer(nn.Module):
         attention_mask: torch.Tensor,
         layer_head_mask: torch.Tensor,
         output_attentions: bool = False,
-    ):
+    ) -> torch.Tensor:
         """
         Args:
             hidden_states (`torch.FloatTensor`): input to the layer of shape `(seq_len, batch, embed_dim)`
             attention_mask (`torch.FloatTensor`): attention mask of size
                 `(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
             layer_head_mask (`torch.FloatTensor`): mask for attention heads in a given layer of size
-                `(config.encoder_attention_heads,)`.
+                `(encoder_attention_heads,)`.
             output_attentions (`bool`, *optional*):
                 Whether or not to return the attentions tensors of all attention layers. See `attentions` under
                 returned tensors for more detail.
@@ -330,7 +330,7 @@ class WhisperEncoderLayer(nn.Module):
         return outputs
 
 
-# Copied from transformers.models.speech_to_text.modeling_speech_to_text.Speech2TextDecoderLayer with Speech2Text->Whisper
+# Copied from transformers.models.mbart.modeling_mbart.MBartDecoderLayer with MBart->Whisper
 class WhisperDecoderLayer(nn.Module):
     def __init__(self, config: WhisperConfig):
         super().__init__()
@@ -369,7 +369,7 @@ class WhisperDecoderLayer(nn.Module):
         past_key_value: Optional[Tuple[torch.Tensor]] = None,
         output_attentions: Optional[bool] = False,
         use_cache: Optional[bool] = True,
-    ):
+    ) -> torch.Tensor:
         """
         Args:
             hidden_states (`torch.FloatTensor`): input to the layer of shape `(seq_len, batch, embed_dim)`
@@ -382,7 +382,7 @@ class WhisperDecoderLayer(nn.Module):
             layer_head_mask (`torch.FloatTensor`): mask for attention heads in a given layer of size
                 `(encoder_attention_heads,)`.
             cross_attn_layer_head_mask (`torch.FloatTensor`): mask for cross-attention heads in a given layer of
-                size *(decoder_attention_heads,)*.
+                size `(decoder_attention_heads,)`.
             past_key_value (`Tuple(torch.FloatTensor)`): cached past key and value projection states
             output_attentions (`bool`, *optional*):
                 Whether or not to return the attentions tensors of all attention layers. See `attentions` under