Improved type hinting for all attention parameters (#28479)

* Changed type hinting for all attention inputs to 'Optional[Tuple[torch.FloatTensor,...]] = None'

* Fixed the ruff formatting issue

* fixed type hinting for all hidden_states to 'Optional[Tuple[torch.FloatTensor, ...]] = None'

* Changed type hinting in these 12 scripts modeling_dpr.py,modeling_nat.py,idefics/vision.py,modeling_tf_dpr.py,modeling_luke.py,modeling_swin.py,modeling_tf_swin.py,modeling_blip.py,modeling_tf_blip.py,modeling_donut_swin.py,modeling_dinat.py,modeling_swinv2.py

* test fail update

* fixed type hinting for these 15 scripts modeling_xlnet.py,modeling_tf_xlnet.py,modeling_led.py,modeling_tf_led.py,modleing_rwkv.py,modeling_dpt.py,modeling_tf_cvt.py,modeling_clip.py,modeling_flax_clip.py,modeling_tf_clip.py,modeling_longformer.py,modeling_tf_longformer.py,modeling_siglip.py,modeling_clap.py,modeling_git.py

* Changed type hinting in these 12 scripts modeling_dpr.py,modeling_nat.py,idefics/vision.py,modeling_tf_dpr.py,modeling_luke.py,modeling_swin.py,modeling_tf_swin.py,modeling_blip.py,modeling_tf_blip.py,modeling_donut_swin.py,modeling_dinat.py,modeling_swinv2.py

* test fail update

* Removed the myvenv file

* Fixed type hinting for these 8 scripts modeling_tvlt.py,modeling_sam.py,modeling_tf_sam.py,modeling_tvp.py,modeling_rag.py,modeling_tf_rag.py,modeling_tf_xlm.py,modeling_xlm.py

src/transformers/models/nat/modeling_nat.py

@@ -104,9 +104,9 @@ class NatEncoderOutput(ModelOutput):
     """
 
     last_hidden_state: torch.FloatTensor = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
-    reshaped_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
+    reshaped_hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 @dataclass
@@ -140,9 +140,9 @@ class NatModelOutput(ModelOutput):
 
     last_hidden_state: torch.FloatTensor = None
     pooler_output: Optional[torch.FloatTensor] = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
-    reshaped_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
+    reshaped_hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 @dataclass
@@ -176,9 +176,9 @@ class NatImageClassifierOutput(ModelOutput):
 
     loss: Optional[torch.FloatTensor] = None
     logits: torch.FloatTensor = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
-    reshaped_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
+    reshaped_hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 class NatEmbeddings(nn.Module):

src/transformers/models/rag/modeling_rag.py

@@ -120,14 +120,14 @@ class RetrievAugLMMarginOutput(ModelOutput):
     context_input_ids: Optional[torch.LongTensor] = None
     context_attention_mask: Optional[torch.LongTensor] = None
     question_encoder_last_hidden_state: Optional[torch.FloatTensor] = None
-    question_enc_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    question_enc_attentions: Optional[Tuple[torch.FloatTensor]] = None
+    question_enc_hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    question_enc_attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
     generator_enc_last_hidden_state: Optional[torch.FloatTensor] = None
-    generator_enc_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    generator_enc_attentions: Optional[Tuple[torch.FloatTensor]] = None
-    generator_dec_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    generator_dec_attentions: Optional[Tuple[torch.FloatTensor]] = None
-    generator_cross_attentions: Optional[Tuple[torch.FloatTensor]] = None
+    generator_enc_hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    generator_enc_attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
+    generator_dec_hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    generator_dec_attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
+    generator_cross_attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 @dataclass
@@ -210,14 +210,14 @@ class RetrievAugLMOutput(ModelOutput):
     context_input_ids: Optional[torch.LongTensor] = None
     context_attention_mask: Optional[torch.LongTensor] = None
     question_encoder_last_hidden_state: Optional[torch.FloatTensor] = None
-    question_enc_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    question_enc_attentions: Optional[Tuple[torch.FloatTensor]] = None
+    question_enc_hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    question_enc_attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
     generator_enc_last_hidden_state: Optional[torch.FloatTensor] = None
-    generator_enc_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    generator_enc_attentions: Optional[Tuple[torch.FloatTensor]] = None
-    generator_dec_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    generator_dec_attentions: Optional[Tuple[torch.FloatTensor]] = None
-    generator_cross_attentions: Optional[Tuple[torch.FloatTensor]] = None
+    generator_enc_hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    generator_enc_attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
+    generator_dec_hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    generator_dec_attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
+    generator_cross_attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 class RagPreTrainedModel(PreTrainedModel):

src/transformers/models/rag/modeling_tf_rag.py

@@ -123,13 +123,13 @@ class TFRetrievAugLMMarginOutput(ModelOutput):
     context_input_ids: tf.Tensor | None = None
     context_attention_mask: tf.Tensor | None = None
     question_encoder_last_hidden_state: tf.Tensor | None = None
-    question_enc_hidden_states: Tuple[tf.Tensor] | None = None
-    question_enc_attentions: Tuple[tf.Tensor] | None = None
+    question_enc_hidden_states: Tuple[tf.Tensor, ...] | None = None
+    question_enc_attentions: Tuple[tf.Tensor, ...] | None = None
     generator_enc_last_hidden_state: tf.Tensor | None = None
-    generator_enc_hidden_states: Tuple[tf.Tensor] | None = None
-    generator_enc_attentions: Tuple[tf.Tensor] | None = None
-    generator_dec_hidden_states: Tuple[tf.Tensor] | None = None
-    generator_dec_attentions: Tuple[tf.Tensor] | None = None
+    generator_enc_hidden_states: Tuple[tf.Tensor, ...] | None = None
+    generator_enc_attentions: Tuple[tf.Tensor, ...] | None = None
+    generator_dec_hidden_states: Tuple[tf.Tensor, ...] | None = None
+    generator_dec_attentions: Tuple[tf.Tensor, ...] | None = None
 
 
 @dataclass
@@ -206,13 +206,13 @@ class TFRetrievAugLMOutput(ModelOutput):
     context_input_ids: tf.Tensor | None = None
     context_attention_mask: tf.Tensor | None = None
     question_encoder_last_hidden_state: tf.Tensor | None = None
-    question_enc_hidden_states: Tuple[tf.Tensor] | None = None
-    question_enc_attentions: Tuple[tf.Tensor] | None = None
+    question_enc_hidden_states: Tuple[tf.Tensor, ...] | None = None
+    question_enc_attentions: Tuple[tf.Tensor, ...] | None = None
     generator_enc_last_hidden_state: tf.Tensor | None = None
-    generator_enc_hidden_states: Tuple[tf.Tensor] | None = None
-    generator_enc_attentions: Tuple[tf.Tensor] | None = None
-    generator_dec_hidden_states: Tuple[tf.Tensor] | None = None
-    generator_dec_attentions: Tuple[tf.Tensor] | None = None
+    generator_enc_hidden_states: Tuple[tf.Tensor, ...] | None = None
+    generator_enc_attentions: Tuple[tf.Tensor, ...] | None = None
+    generator_dec_hidden_states: Tuple[tf.Tensor, ...] | None = None
+    generator_dec_attentions: Tuple[tf.Tensor, ...] | None = None
 
 
 class TFRagPreTrainedModel(TFPreTrainedModel):

src/transformers/models/rwkv/modeling_rwkv.py

@@ -493,8 +493,8 @@ class RwkvOutput(ModelOutput):
 
     last_hidden_state: torch.FloatTensor = None
     state: Optional[List[torch.FloatTensor]] = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 @dataclass
@@ -526,8 +526,8 @@ class RwkvCausalLMOutput(ModelOutput):
     loss: Optional[torch.FloatTensor] = None
     logits: torch.FloatTensor = None
     state: Optional[List[torch.FloatTensor]] = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 RWKV_START_DOCSTRING = r"""

src/transformers/models/sam/modeling_sam.py

@@ -71,8 +71,8 @@ class SamVisionEncoderOutput(ModelOutput):
 
     image_embeds: Optional[torch.FloatTensor] = None
     last_hidden_state: torch.FloatTensor = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 @dataclass
@@ -106,9 +106,9 @@ class SamImageSegmentationOutput(ModelOutput):
 
     iou_scores: torch.FloatTensor = None
     pred_masks: torch.FloatTensor = None
-    vision_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    vision_attentions: Optional[Tuple[torch.FloatTensor]] = None
-    mask_decoder_attentions: Optional[Tuple[torch.FloatTensor]] = None
+    vision_hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    vision_attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
+    mask_decoder_attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 class SamPatchEmbeddings(nn.Module):

src/transformers/models/sam/modeling_tf_sam.py

@@ -74,8 +74,8 @@ class TFSamVisionEncoderOutput(ModelOutput):
 
     image_embeds: tf.Tensor | None = None
     last_hidden_state: tf.Tensor = None
-    hidden_states: Tuple[tf.Tensor] | None = None
-    attentions: Tuple[tf.Tensor] | None = None
+    hidden_states: Tuple[tf.Tensor, ...] | None = None
+    attentions: Tuple[tf.Tensor, ...] | None = None
 
 
 @dataclass
@@ -109,9 +109,9 @@ class TFSamImageSegmentationOutput(ModelOutput):
 
     iou_scores: tf.Tensor = None
     pred_masks: tf.Tensor = None
-    vision_hidden_states: Tuple[tf.Tensor] | None = None
-    vision_attentions: Tuple[tf.Tensor] | None = None
-    mask_decoder_attentions: Tuple[tf.Tensor] | None = None
+    vision_hidden_states: Tuple[tf.Tensor, ...] | None = None
+    vision_attentions: Tuple[tf.Tensor, ...] | None = None
+    mask_decoder_attentions: Tuple[tf.Tensor, ...] | None = None
 
 
 class TFSamPatchEmbeddings(tf.keras.layers.Layer):

src/transformers/models/siglip/modeling_siglip.py

@@ -171,8 +171,8 @@ class SiglipVisionModelOutput(ModelOutput):
 
     image_embeds: Optional[torch.FloatTensor] = None
     last_hidden_state: torch.FloatTensor = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 @dataclass
@@ -201,8 +201,8 @@ class SiglipTextModelOutput(ModelOutput):
 
     text_embeds: Optional[torch.FloatTensor] = None
     last_hidden_state: torch.FloatTensor = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 @dataclass

src/transformers/models/swin/modeling_swin.py

@@ -92,9 +92,9 @@ class SwinEncoderOutput(ModelOutput):
     """
 
     last_hidden_state: torch.FloatTensor = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
-    reshaped_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
+    reshaped_hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 @dataclass
@@ -128,9 +128,9 @@ class SwinModelOutput(ModelOutput):
 
     last_hidden_state: torch.FloatTensor = None
     pooler_output: Optional[torch.FloatTensor] = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
-    reshaped_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
+    reshaped_hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 @dataclass
@@ -164,9 +164,9 @@ class SwinMaskedImageModelingOutput(ModelOutput):
 
     loss: Optional[torch.FloatTensor] = None
     reconstruction: torch.FloatTensor = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
-    reshaped_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
+    reshaped_hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
 
     @property
     def logits(self):
@@ -209,9 +209,9 @@ class SwinImageClassifierOutput(ModelOutput):
 
     loss: Optional[torch.FloatTensor] = None
     logits: torch.FloatTensor = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
-    reshaped_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
+    reshaped_hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 def window_partition(input_feature, window_size):

src/transformers/models/swin/modeling_tf_swin.py

@@ -97,9 +97,9 @@ class TFSwinEncoderOutput(ModelOutput):
     """
 
     last_hidden_state: tf.Tensor = None
-    hidden_states: Tuple[tf.Tensor] | None = None
-    attentions: Tuple[tf.Tensor] | None = None
-    reshaped_hidden_states: Tuple[tf.Tensor] | None = None
+    hidden_states: Tuple[tf.Tensor, ...] | None = None
+    attentions: Tuple[tf.Tensor, ...] | None = None
+    reshaped_hidden_states: Tuple[tf.Tensor, ...] | None = None
 
 
 @dataclass
@@ -133,9 +133,9 @@ class TFSwinModelOutput(ModelOutput):
 
     last_hidden_state: tf.Tensor = None
     pooler_output: tf.Tensor | None = None
-    hidden_states: Tuple[tf.Tensor] | None = None
-    attentions: Tuple[tf.Tensor] | None = None
-    reshaped_hidden_states: Tuple[tf.Tensor] | None = None
+    hidden_states: Tuple[tf.Tensor, ...] | None = None
+    attentions: Tuple[tf.Tensor, ...] | None = None
+    reshaped_hidden_states: Tuple[tf.Tensor, ...] | None = None
 
 
 @dataclass
@@ -169,9 +169,9 @@ class TFSwinMaskedImageModelingOutput(ModelOutput):
 
     loss: tf.Tensor | None = None
     reconstruction: tf.Tensor = None
-    hidden_states: Tuple[tf.Tensor] | None = None
-    attentions: Tuple[tf.Tensor] | None = None
-    reshaped_hidden_states: Tuple[tf.Tensor] | None = None
+    hidden_states: Tuple[tf.Tensor, ...] | None = None
+    attentions: Tuple[tf.Tensor, ...] | None = None
+    reshaped_hidden_states: Tuple[tf.Tensor, ...] | None = None
 
     @property
     def logits(self):
@@ -214,9 +214,9 @@ class TFSwinImageClassifierOutput(ModelOutput):
 
     loss: tf.Tensor | None = None
     logits: tf.Tensor = None
-    hidden_states: Tuple[tf.Tensor] | None = None
-    attentions: Tuple[tf.Tensor] | None = None
-    reshaped_hidden_states: Tuple[tf.Tensor] | None = None
+    hidden_states: Tuple[tf.Tensor, ...] | None = None
+    attentions: Tuple[tf.Tensor, ...] | None = None
+    reshaped_hidden_states: Tuple[tf.Tensor, ...] | None = None
 
 
 def window_partition(input_feature: tf.Tensor, window_size: int) -> tf.Tensor:

src/transformers/models/swinv2/modeling_swinv2.py

@@ -94,9 +94,9 @@ class Swinv2EncoderOutput(ModelOutput):
     """
 
     last_hidden_state: torch.FloatTensor = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
-    reshaped_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
+    reshaped_hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 @dataclass
@@ -131,9 +131,9 @@ class Swinv2ModelOutput(ModelOutput):
 
     last_hidden_state: torch.FloatTensor = None
     pooler_output: Optional[torch.FloatTensor] = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
-    reshaped_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
+    reshaped_hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 @dataclass
@@ -168,9 +168,9 @@ class Swinv2MaskedImageModelingOutput(ModelOutput):
 
     loss: Optional[torch.FloatTensor] = None
     reconstruction: torch.FloatTensor = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
-    reshaped_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
+    reshaped_hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
 
     @property
     def logits(self):
@@ -214,9 +214,9 @@ class Swinv2ImageClassifierOutput(ModelOutput):
 
     loss: Optional[torch.FloatTensor] = None
     logits: torch.FloatTensor = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
-    reshaped_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
+    reshaped_hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 # Copied from transformers.models.swin.modeling_swin.window_partition

src/transformers/models/tvlt/modeling_tvlt.py

@@ -88,8 +88,8 @@ class TvltModelOutput(ModelOutput):
     audio_label_masks: torch.LongTensor = None
     pixel_ids_restore: torch.LongTensor = None
     audio_ids_restore: torch.LongTensor = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 @dataclass
@@ -111,8 +111,8 @@ class TvltDecoderOutput(ModelOutput):
     """
 
     logits: torch.FloatTensor = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 @dataclass
@@ -145,8 +145,8 @@ class TvltForPreTrainingOutput(ModelOutput):
     matching_logits: torch.FloatTensor = None
     pixel_logits: torch.FloatTensor = None
     audio_logits: torch.FloatTensor = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 def generate_pixel_mask_noise(pixel_values, pixel_mask=None, mask_ratio=0.75):

src/transformers/models/tvp/modeling_tvp.py

@@ -61,8 +61,8 @@ class TvpVideoGroundingOutput(ModelOutput):
 
     loss: Optional[torch.FloatTensor] = None
     logits: torch.FloatTensor = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 class TvpLoss(nn.Module):

src/transformers/models/xlm/modeling_tf_xlm.py

@@ -614,8 +614,8 @@ class TFXLMWithLMHeadModelOutput(ModelOutput):
     """
 
     logits: tf.Tensor = None
-    hidden_states: Tuple[tf.Tensor] | None = None
-    attentions: Tuple[tf.Tensor] | None = None
+    hidden_states: Tuple[tf.Tensor, ...] | None = None
+    attentions: Tuple[tf.Tensor, ...] | None = None
 
 
 XLM_START_DOCSTRING = r"""

src/transformers/models/xlm/modeling_xlm.py

@@ -297,8 +297,8 @@ class XLMForQuestionAnsweringOutput(ModelOutput):
     end_top_log_probs: Optional[torch.FloatTensor] = None
     end_top_index: Optional[torch.LongTensor] = None
     cls_logits: Optional[torch.FloatTensor] = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 XLM_START_DOCSTRING = r"""

src/transformers/models/xlnet/modeling_tf_xlnet.py

@@ -871,8 +871,8 @@ class TFXLNetModelOutput(ModelOutput):
 
     last_hidden_state: tf.Tensor = None
     mems: List[tf.Tensor] | None = None
-    hidden_states: Tuple[tf.Tensor] | None = None
-    attentions: Tuple[tf.Tensor] | None = None
+    hidden_states: Tuple[tf.Tensor, ...] | None = None
+    attentions: Tuple[tf.Tensor, ...] | None = None
 
 
 @dataclass
@@ -908,8 +908,8 @@ class TFXLNetLMHeadModelOutput(ModelOutput):
     loss: tf.Tensor | None = None
     logits: tf.Tensor = None
     mems: List[tf.Tensor] | None = None
-    hidden_states: Tuple[tf.Tensor] | None = None
-    attentions: Tuple[tf.Tensor] | None = None
+    hidden_states: Tuple[tf.Tensor, ...] | None = None
+    attentions: Tuple[tf.Tensor, ...] | None = None
 
 
 @dataclass
@@ -942,8 +942,8 @@ class TFXLNetForSequenceClassificationOutput(ModelOutput):
     loss: tf.Tensor | None = None
     logits: tf.Tensor = None
     mems: List[tf.Tensor] | None = None
-    hidden_states: Tuple[tf.Tensor] | None = None
-    attentions: Tuple[tf.Tensor] | None = None
+    hidden_states: Tuple[tf.Tensor, ...] | None = None
+    attentions: Tuple[tf.Tensor, ...] | None = None
 
 
 @dataclass
@@ -976,8 +976,8 @@ class TFXLNetForTokenClassificationOutput(ModelOutput):
     loss: tf.Tensor | None = None
     logits: tf.Tensor = None
     mems: List[tf.Tensor] | None = None
-    hidden_states: Tuple[tf.Tensor] | None = None
-    attentions: Tuple[tf.Tensor] | None = None
+    hidden_states: Tuple[tf.Tensor, ...] | None = None
+    attentions: Tuple[tf.Tensor, ...] | None = None
 
 
 @dataclass
@@ -1012,8 +1012,8 @@ class TFXLNetForMultipleChoiceOutput(ModelOutput):
     loss: tf.Tensor | None = None
     logits: tf.Tensor = None
     mems: List[tf.Tensor] | None = None
-    hidden_states: Tuple[tf.Tensor] | None = None
-    attentions: Tuple[tf.Tensor] | None = None
+    hidden_states: Tuple[tf.Tensor, ...] | None = None
+    attentions: Tuple[tf.Tensor, ...] | None = None
 
 
 @dataclass
@@ -1049,8 +1049,8 @@ class TFXLNetForQuestionAnsweringSimpleOutput(ModelOutput):
     start_logits: tf.Tensor = None
     end_logits: tf.Tensor = None
     mems: List[tf.Tensor] | None = None
-    hidden_states: Tuple[tf.Tensor] | None = None
-    attentions: Tuple[tf.Tensor] | None = None
+    hidden_states: Tuple[tf.Tensor, ...] | None = None
+    attentions: Tuple[tf.Tensor, ...] | None = None
 
 
 XLNET_START_DOCSTRING = r"""

src/transformers/models/xlnet/modeling_xlnet.py

@@ -605,8 +605,8 @@ class XLNetModelOutput(ModelOutput):
 
     last_hidden_state: torch.FloatTensor
     mems: Optional[List[torch.FloatTensor]] = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 @dataclass
@@ -642,8 +642,8 @@ class XLNetLMHeadModelOutput(ModelOutput):
     loss: Optional[torch.FloatTensor] = None
     logits: torch.FloatTensor = None
     mems: Optional[List[torch.FloatTensor]] = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 @dataclass
@@ -676,8 +676,8 @@ class XLNetForSequenceClassificationOutput(ModelOutput):
     loss: Optional[torch.FloatTensor] = None
     logits: torch.FloatTensor = None
     mems: Optional[List[torch.FloatTensor]] = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 @dataclass
@@ -710,8 +710,8 @@ class XLNetForTokenClassificationOutput(ModelOutput):
     loss: Optional[torch.FloatTensor] = None
     logits: torch.FloatTensor = None
     mems: Optional[List[torch.FloatTensor]] = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 @dataclass
@@ -746,8 +746,8 @@ class XLNetForMultipleChoiceOutput(ModelOutput):
     loss: Optional[torch.FloatTensor] = None
     logits: torch.FloatTensor = None
     mems: Optional[List[torch.FloatTensor]] = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 @dataclass
@@ -783,8 +783,8 @@ class XLNetForQuestionAnsweringSimpleOutput(ModelOutput):
     start_logits: torch.FloatTensor = None
     end_logits: torch.FloatTensor = None
     mems: Optional[List[torch.FloatTensor]] = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 @dataclass
@@ -831,8 +831,8 @@ class XLNetForQuestionAnsweringOutput(ModelOutput):
     end_top_index: Optional[torch.LongTensor] = None
     cls_logits: Optional[torch.FloatTensor] = None
     mems: Optional[List[torch.FloatTensor]] = None
-    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
-    attentions: Optional[Tuple[torch.FloatTensor]] = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
 
 
 XLNET_START_DOCSTRING = r"""