diff --git a/src/transformers/models/mpnet/modeling_tf_mpnet.py b/src/transformers/models/mpnet/modeling_tf_mpnet.py
index 3fc43184617..3ceb1489a9b 100644
--- a/src/transformers/models/mpnet/modeling_tf_mpnet.py
+++ b/src/transformers/models/mpnet/modeling_tf_mpnet.py
@@ -18,7 +18,9 @@
 
 import math
 import warnings
+from typing import Optional, Tuple, Union
 
+import numpy as np
 import tensorflow as tf
 
 from ...activations_tf import get_tf_activation
@@ -33,6 +35,7 @@ from ...modeling_tf_outputs import (
 )
 from ...modeling_tf_utils import (
     TFMaskedLanguageModelingLoss,
+    TFModelInputType,
     TFMultipleChoiceLoss,
     TFPreTrainedModel,
     TFQuestionAnsweringLoss,
@@ -681,16 +684,16 @@ class TFMPNetModel(TFMPNetPreTrainedModel):
     )
     def call(
         self,
-        input_ids=None,
-        attention_mask=None,
-        position_ids=None,
-        head_mask=None,
-        inputs_embeds=None,
-        output_attentions=None,
-        output_hidden_states=None,
-        return_dict=None,
-        training=False,
-    ):
+        input_ids: Optional[TFModelInputType] = None,
+        attention_mask: Optional[Union[np.array, tf.Tensor]] = None,
+        position_ids: Optional[Union[np.array, tf.Tensor]] = None,
+        head_mask: Optional[Union[np.array, tf.Tensor]] = None,
+        inputs_embeds: Optional[tf.Tensor] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+        training: bool = False,
+    ) -> Union[TFBaseModelOutput, Tuple[tf.Tensor]]:
         outputs = self.mpnet(
             input_ids=input_ids,
             attention_mask=attention_mask,
@@ -796,17 +799,17 @@ class TFMPNetForMaskedLM(TFMPNetPreTrainedModel, TFMaskedLanguageModelingLoss):
     )
     def call(
         self,
-        input_ids=None,
-        attention_mask=None,
-        position_ids=None,
-        head_mask=None,
-        inputs_embeds=None,
-        output_attentions=None,
-        output_hidden_states=None,
-        return_dict=None,
-        labels=None,
-        training=False,
-    ):
+        input_ids: Optional[TFModelInputType] = None,
+        attention_mask: Optional[Union[np.ndarray, tf.Tensor]] = None,
+        position_ids: Optional[Union[np.ndarray, tf.Tensor]] = None,
+        head_mask: Optional[Union[np.ndarray, tf.Tensor]] = None,
+        inputs_embeds: Optional[tf.Tensor] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+        labels: Optional[tf.Tensor] = None,
+        training: bool = False,
+    ) -> Union[TFMaskedLMOutput, Tuple[tf.Tensor]]:
         r"""
         labels (`tf.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
             Labels for computing the masked language modeling loss. Indices should be in `[-100, 0, ...,
@@ -901,17 +904,17 @@ class TFMPNetForSequenceClassification(TFMPNetPreTrainedModel, TFSequenceClassif
     )
     def call(
         self,
-        input_ids=None,
-        attention_mask=None,
-        position_ids=None,
-        head_mask=None,
-        inputs_embeds=None,
-        output_attentions=None,
-        output_hidden_states=None,
-        return_dict=None,
-        labels=None,
-        training=False,
-    ):
+        input_ids: Optional[TFModelInputType] = None,
+        attention_mask: Optional[Union[np.array, tf.Tensor]] = None,
+        position_ids: Optional[Union[np.array, tf.Tensor]] = None,
+        head_mask: Optional[Union[np.array, tf.Tensor]] = None,
+        inputs_embeds: Optional[tf.Tensor] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+        labels: Optional[tf.Tensor] = None,
+        training: bool = False,
+    ) -> Union[TFSequenceClassifierOutput, Tuple[tf.Tensor]]:
         r"""
         labels (`tf.Tensor` of shape `(batch_size,)`, *optional*):
             Labels for computing the sequence classification/regression loss. Indices should be in `[0, ...,
@@ -991,17 +994,17 @@ class TFMPNetForMultipleChoice(TFMPNetPreTrainedModel, TFMultipleChoiceLoss):
     )
     def call(
         self,
-        input_ids=None,
-        attention_mask=None,
-        position_ids=None,
-        head_mask=None,
-        inputs_embeds=None,
-        output_attentions=None,
-        output_hidden_states=None,
-        return_dict=None,
-        labels=None,
-        training=False,
-    ):
+        input_ids: Optional[TFModelInputType] = None,
+        attention_mask: Optional[Union[np.ndarray, tf.Tensor]] = None,
+        position_ids: Optional[Union[np.ndarray, tf.Tensor]] = None,
+        head_mask: Optional[Union[np.ndarray, tf.Tensor]] = None,
+        inputs_embeds: Optional[tf.Tensor] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+        labels: Optional[tf.Tensor] = None,
+        training: bool = False,
+    ) -> Union[TFMultipleChoiceModelOutput, Tuple[tf.Tensor]]:
         r"""
         labels (`tf.Tensor` of shape `(batch_size,)`, *optional*):
             Labels for computing the multiple choice classification loss. Indices should be in `[0, ..., num_choices]`
@@ -1102,17 +1105,17 @@ class TFMPNetForTokenClassification(TFMPNetPreTrainedModel, TFTokenClassificatio
     )
     def call(
         self,
-        input_ids=None,
-        attention_mask=None,
-        position_ids=None,
-        head_mask=None,
-        inputs_embeds=None,
-        output_attentions=None,
-        output_hidden_states=None,
-        return_dict=None,
-        labels=None,
-        training=False,
-    ):
+        input_ids: Optional[TFModelInputType] = None,
+        attention_mask: Optional[Union[np.ndarray, tf.Tensor]] = None,
+        position_ids: Optional[Union[np.ndarray, tf.Tensor]] = None,
+        head_mask: Optional[Union[np.ndarray, tf.Tensor]] = None,
+        inputs_embeds: Optional[tf.Tensor] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+        labels: Optional[tf.Tensor] = None,
+        training: bool = False,
+    ) -> Union[TFTokenClassifierOutput, Tuple[tf.Tensor]]:
         r"""
         labels (`tf.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
             Labels for computing the token classification loss. Indices should be in `[0, ..., config.num_labels - 1]`.
@@ -1184,19 +1187,19 @@ class TFMPNetForQuestionAnswering(TFMPNetPreTrainedModel, TFQuestionAnsweringLos
     )
     def call(
         self,
-        input_ids=None,
-        attention_mask=None,
-        position_ids=None,
-        head_mask=None,
-        inputs_embeds=None,
-        output_attentions=None,
-        output_hidden_states=None,
-        return_dict=None,
-        start_positions=None,
-        end_positions=None,
-        training=False,
+        input_ids: Optional[TFModelInputType] = None,
+        attention_mask: Optional[Union[np.array, tf.Tensor]] = None,
+        position_ids: Optional[Union[np.array, tf.Tensor]] = None,
+        head_mask: Optional[Union[np.array, tf.Tensor]] = None,
+        inputs_embeds: Optional[tf.Tensor] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+        start_positions: Optional[tf.Tensor] = None,
+        end_positions: Optional[tf.Tensor] = None,
+        training: bool = False,
         **kwargs,
-    ):
+    ) -> Union[TFQuestionAnsweringModelOutput, Tuple[tf.Tensor]]:
         r"""
         start_positions (`tf.Tensor` of shape `(batch_size,)`, *optional*):
             Labels for position (index) of the start of the labelled span for computing the token classification loss.