Add type hints for PyTorch UniSpeech, MPNet and Nystromformer (#19039)

* added type hints pytorch unispeech

* added type hints pytorch  MPNet

* added type hints nystromformer

* resolved copy inconsistencies

* make fix-copies
Co-authored-by: matt <rocketknight1@gmail.com>

src/transformers/models/data2vec/modeling_data2vec_audio.py

@@ -563,11 +563,11 @@ class Data2VecAudioEncoder(nn.Module):
 
     def forward(
         self,
-        hidden_states,
-        attention_mask=None,
-        output_attentions=False,
-        output_hidden_states=False,
-        return_dict=True,
+        hidden_states: torch.tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        output_attentions: bool = False,
+        output_hidden_states: bool = False,
+        return_dict: bool = True,
     ):
         all_hidden_states = () if output_hidden_states else None
         all_self_attentions = () if output_attentions else None

src/transformers/models/hubert/modeling_hubert.py

@@ -618,7 +618,12 @@ class HubertEncoderLayerStableLayerNorm(nn.Module):
         self.feed_forward = HubertFeedForward(config)
         self.final_layer_norm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps)
 
-    def forward(self, hidden_states, attention_mask=None, output_attentions=False):
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        output_attentions: bool = False,
+    ):
         attn_residual = hidden_states
         hidden_states = self.layer_norm(hidden_states)
         hidden_states, attn_weights, _ = self.attention(
@@ -649,11 +654,11 @@ class HubertEncoder(nn.Module):
 
     def forward(
         self,
-        hidden_states,
-        attention_mask=None,
-        output_attentions=False,
-        output_hidden_states=False,
-        return_dict=True,
+        hidden_states: torch.tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        output_attentions: bool = False,
+        output_hidden_states: bool = False,
+        return_dict: bool = True,
     ):
         all_hidden_states = () if output_hidden_states else None
         all_self_attentions = () if output_attentions else None

src/transformers/models/mpnet/modeling_mpnet.py

@@ -323,12 +323,12 @@ class MPNetEncoder(nn.Module):
 
     def forward(
         self,
-        hidden_states,
-        attention_mask=None,
-        head_mask=None,
-        output_attentions=False,
-        output_hidden_states=False,
-        return_dict=False,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        head_mask: Optional[torch.Tensor] = None,
+        output_attentions: bool = False,
+        output_hidden_states: bool = False,
+        return_dict: bool = False,
         **kwargs,
     ):
         position_bias = self.compute_position_bias(hidden_states)

src/transformers/models/nystromformer/modeling_nystromformer.py

@@ -354,12 +354,12 @@ class NystromformerEncoder(nn.Module):
 
     def forward(
         self,
-        hidden_states,
-        attention_mask=None,
-        head_mask=None,
-        output_attentions=False,
-        output_hidden_states=False,
-        return_dict=True,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        head_mask: Optional[torch.Tensor] = None,
+        output_attentions: bool = False,
+        output_hidden_states: bool = False,
+        return_dict: bool = True,
     ):
         all_hidden_states = () if output_hidden_states else None
         all_self_attentions = () if output_attentions else None

src/transformers/models/unispeech/modeling_unispeech.py

@@ -655,7 +655,12 @@ class UniSpeechEncoderLayerStableLayerNorm(nn.Module):
         self.feed_forward = UniSpeechFeedForward(config)
         self.final_layer_norm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps)
 
-    def forward(self, hidden_states, attention_mask=None, output_attentions=False):
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        output_attentions: bool = False,
+    ):
         attn_residual = hidden_states
         hidden_states = self.layer_norm(hidden_states)
         hidden_states, attn_weights, _ = self.attention(
@@ -686,11 +691,11 @@ class UniSpeechEncoder(nn.Module):
 
     def forward(
         self,
-        hidden_states,
-        attention_mask=None,
-        output_attentions=False,
-        output_hidden_states=False,
-        return_dict=True,
+        hidden_states: torch.tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        output_attentions: bool = False,
+        output_hidden_states: bool = False,
+        return_dict: bool = True,
     ):
         all_hidden_states = () if output_hidden_states else None
         all_self_attentions = () if output_attentions else None

src/transformers/models/unispeech_sat/modeling_unispeech_sat.py

@@ -669,7 +669,12 @@ class UniSpeechSatEncoderLayerStableLayerNorm(nn.Module):
         self.feed_forward = UniSpeechSatFeedForward(config)
         self.final_layer_norm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps)
 
-    def forward(self, hidden_states, attention_mask=None, output_attentions=False):
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        output_attentions: bool = False,
+    ):
         attn_residual = hidden_states
         hidden_states = self.layer_norm(hidden_states)
         hidden_states, attn_weights, _ = self.attention(
@@ -700,11 +705,11 @@ class UniSpeechSatEncoder(nn.Module):
 
     def forward(
         self,
-        hidden_states,
-        attention_mask=None,
-        output_attentions=False,
-        output_hidden_states=False,
-        return_dict=True,
+        hidden_states: torch.tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        output_attentions: bool = False,
+        output_hidden_states: bool = False,
+        return_dict: bool = True,
     ):
         all_hidden_states = () if output_hidden_states else None
         all_self_attentions = () if output_attentions else None

src/transformers/models/wav2vec2/modeling_wav2vec2.py

@@ -704,7 +704,12 @@ class Wav2Vec2EncoderLayerStableLayerNorm(nn.Module):
         self.feed_forward = Wav2Vec2FeedForward(config)
         self.final_layer_norm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps)
 
-    def forward(self, hidden_states, attention_mask=None, output_attentions=False):
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        output_attentions: bool = False,
+    ):
         attn_residual = hidden_states
         hidden_states = self.layer_norm(hidden_states)
         hidden_states, attn_weights, _ = self.attention(
@@ -734,11 +739,11 @@ class Wav2Vec2Encoder(nn.Module):
 
     def forward(
         self,
-        hidden_states,
-        attention_mask=None,
-        output_attentions=False,
-        output_hidden_states=False,
-        return_dict=True,
+        hidden_states: torch.tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        output_attentions: bool = False,
+        output_hidden_states: bool = False,
+        return_dict: bool = True,
     ):
         all_hidden_states = () if output_hidden_states else None
         all_self_attentions = () if output_attentions else None