Post-PR fixes! (#38868)

* Post-PR fixes!

* make fix-copies

src/transformers/models/lightglue/convert_lightglue_to_hf.py

@@ -15,7 +15,6 @@ import argparse
 import gc
 import os
 import re
-from typing import List
 
 import torch
 from datasets import load_dataset
@@ -90,7 +89,7 @@ ORIGINAL_TO_CONVERTED_KEY_MAPPING = {
 }
 
 
-def convert_old_keys_to_new_keys(state_dict_keys: List[str]):
+def convert_old_keys_to_new_keys(state_dict_keys: list[str]):
     """
     This function should be applied only once, on the concatenated keys to efficiently rename using
     the key mappings.

src/transformers/models/lightglue/image_processing_lightglue.py

@@ -17,7 +17,7 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-from typing import Dict, List, Optional, Tuple, Union
+from typing import Optional, Union
 
 import numpy as np
 import torch
@@ -139,7 +139,7 @@ class LightGlueImageProcessor(BaseImageProcessor):
         do_resize (`bool`, *optional*, defaults to `True`):
             Controls whether to resize the image's (height, width) dimensions to the specified `size`. Can be overridden
             by `do_resize` in the `preprocess` method.
-        size (`Dict[str, int]` *optional*, defaults to `{"height": 480, "width": 640}`):
+        size (`dict[str, int]` *optional*, defaults to `{"height": 480, "width": 640}`):
             Resolution of the output image after `resize` is applied. Only has an effect if `do_resize` is set to
             `True`. Can be overridden by `size` in the `preprocess` method.
         resample (`PILImageResampling`, *optional*, defaults to `Resampling.BILINEAR`):
@@ -159,7 +159,7 @@ class LightGlueImageProcessor(BaseImageProcessor):
     def __init__(
         self,
         do_resize: bool = True,
-        size: Optional[Dict[str, int]] = None,
+        size: Optional[dict[str, int]] = None,
         resample: PILImageResampling = PILImageResampling.BILINEAR,
         do_rescale: bool = True,
         rescale_factor: float = 1 / 255,
@@ -180,7 +180,7 @@ class LightGlueImageProcessor(BaseImageProcessor):
     def resize(
         self,
         image: np.ndarray,
-        size: Dict[str, int],
+        size: dict[str, int],
         data_format: Optional[Union[str, ChannelDimension]] = None,
         input_data_format: Optional[Union[str, ChannelDimension]] = None,
         **kwargs,
@@ -191,7 +191,7 @@ class LightGlueImageProcessor(BaseImageProcessor):
         Args:
             image (`np.ndarray`):
                 Image to resize.
-            size (`Dict[str, int]`):
+            size (`dict[str, int]`):
                 Dictionary of the form `{"height": int, "width": int}`, specifying the size of the output image.
             data_format (`ChannelDimension` or `str`, *optional*):
                 The channel dimension format of the output image. If not provided, it will be inferred from the input
@@ -220,7 +220,7 @@ class LightGlueImageProcessor(BaseImageProcessor):
         self,
         images,
         do_resize: Optional[bool] = None,
-        size: Optional[Dict[str, int]] = None,
+        size: Optional[dict[str, int]] = None,
         resample: PILImageResampling = None,
         do_rescale: Optional[bool] = None,
         rescale_factor: Optional[float] = None,
@@ -240,7 +240,7 @@ class LightGlueImageProcessor(BaseImageProcessor):
                 `do_rescale=False`.
             do_resize (`bool`, *optional*, defaults to `self.do_resize`):
                 Whether to resize the image.
-            size (`Dict[str, int]`, *optional*, defaults to `self.size`):
+            size (`dict[str, int]`, *optional*, defaults to `self.size`):
                 Size of the output image after `resize` has been applied. If `size["shortest_edge"]` >= 384, the image
                 is resized to `(size["shortest_edge"], size["shortest_edge"])`. Otherwise, the smaller edge of the
                 image will be matched to `int(size["shortest_edge"]/ crop_pct)`, after which the image is cropped to
@@ -337,23 +337,23 @@ class LightGlueImageProcessor(BaseImageProcessor):
     def post_process_keypoint_matching(
         self,
         outputs: LightGlueKeypointMatchingOutput,
-        target_sizes: Union[TensorType, List[Tuple]],
+        target_sizes: Union[TensorType, list[tuple]],
         threshold: float = 0.0,
-    ) -> List[Dict[str, torch.Tensor]]:
+    ) -> list[dict[str, torch.Tensor]]:
         """
         Converts the raw output of [`KeypointMatchingOutput`] into lists of keypoints, scores and descriptors
         with coordinates absolute to the original image sizes.
         Args:
             outputs ([`KeypointMatchingOutput`]):
                 Raw outputs of the model.
-            target_sizes (`torch.Tensor` or `List[Tuple[Tuple[int, int]]]`, *optional*):
-                Tensor of shape `(batch_size, 2, 2)` or list of tuples of tuples (`Tuple[int, int]`) containing the
+            target_sizes (`torch.Tensor` or `list[tuple[tuple[int, int]]]`, *optional*):
+                Tensor of shape `(batch_size, 2, 2)` or list of tuples of tuples (`tuple[int, int]`) containing the
                 target size `(height, width)` of each image in the batch. This must be the original image size (before
                 any processing).
             threshold (`float`, *optional*, defaults to 0.0):
                 Threshold to filter out the matches with low scores.
         Returns:
-            `List[Dict]`: A list of dictionaries, each dictionary containing the keypoints in the first and second image
+            `list[Dict]`: A list of dictionaries, each dictionary containing the keypoints in the first and second image
             of the pair, the matching scores and the matching indices.
         """
         if outputs.mask.shape[0] != len(target_sizes):
@@ -361,7 +361,7 @@ class LightGlueImageProcessor(BaseImageProcessor):
         if not all(len(target_size) == 2 for target_size in target_sizes):
             raise ValueError("Each element of target_sizes must contain the size (h, w) of each image of the batch")
 
-        if isinstance(target_sizes, List):
+        if isinstance(target_sizes, list):
             image_pair_sizes = torch.tensor(target_sizes, device=outputs.mask.device)
         else:
             if target_sizes.shape[1] != 2 or target_sizes.shape[2] != 2:

src/transformers/models/lightglue/modeling_lightglue.py

@@ -18,7 +18,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 from dataclasses import dataclass
-from typing import Callable, Optional, Tuple, Union
+from typing import Callable, Optional, Union
 
 import numpy as np
 import torch
@@ -74,8 +74,8 @@ class LightGlueKeypointMatchingOutput(ModelOutput):
     keypoints: Optional[torch.FloatTensor] = None
     prune: Optional[torch.IntTensor] = None
     mask: 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
 
 
 class LightGluePositionalEncoder(nn.Module):
@@ -85,7 +85,7 @@ class LightGluePositionalEncoder(nn.Module):
 
     def forward(
         self, keypoints: torch.Tensor, output_hidden_states: Optional[bool] = False
-    ) -> Union[Tuple[torch.Tensor], Tuple[torch.Tensor, torch.Tensor]]:
+    ) -> Union[tuple[torch.Tensor], tuple[torch.Tensor, torch.Tensor]]:
         projected_keypoints = self.projector(keypoints)
         embeddings = projected_keypoints.repeat_interleave(2, dim=-1)
         cosines = torch.cos(embeddings)
@@ -200,12 +200,12 @@ class LightGlueAttention(nn.Module):
     def forward(
         self,
         hidden_states: torch.Tensor,
-        position_embeddings: Optional[Tuple[torch.Tensor, torch.Tensor]] = None,
+        position_embeddings: Optional[tuple[torch.Tensor, torch.Tensor]] = None,
         attention_mask: Optional[torch.Tensor] = None,
         encoder_hidden_states: Optional[torch.Tensor] = None,
         encoder_attention_mask: Optional[torch.Tensor] = None,
         **kwargs: Unpack[FlashAttentionKwargs],
-    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+    ) -> tuple[torch.Tensor, Optional[torch.Tensor], Optional[tuple[torch.Tensor]]]:
         input_shape = hidden_states.shape[:-1]
         hidden_shape = (*input_shape, -1, self.head_dim)
 
@@ -274,7 +274,7 @@ class LightGlueTransformerLayer(nn.Module):
         attention_mask: torch.Tensor,
         output_hidden_states: Optional[bool] = False,
         output_attentions: Optional[bool] = False,
-    ) -> Tuple[torch.Tensor, Optional[Tuple[torch.Tensor]], Optional[Tuple[torch.Tensor]]]:
+    ) -> tuple[torch.Tensor, Optional[tuple[torch.Tensor]], Optional[tuple[torch.Tensor]]]:
         all_hidden_states = () if output_hidden_states else None
         all_attentions = () if output_attentions else None
 
@@ -435,7 +435,7 @@ class LightGluePreTrainedModel(PreTrainedModel):
             module.weight.data.fill_(1.0)
 
 
-def get_matches_from_scores(scores: torch.Tensor, threshold: float) -> Tuple[torch.Tensor, torch.Tensor]:
+def get_matches_from_scores(scores: torch.Tensor, threshold: float) -> tuple[torch.Tensor, torch.Tensor]:
     """obtain matches from a score matrix [Bx M+1 x N+1]"""
     batch_size, _, _ = scores.shape
     # For each keypoint, get the best match
@@ -548,7 +548,7 @@ class LightGlueForKeypointMatching(LightGluePreTrainedModel):
 
     def _keypoint_processing(
         self, descriptors: torch.Tensor, keypoints: torch.Tensor, output_hidden_states: Optional[bool] = False
-    ) -> Tuple[torch.Tensor, Tuple[torch.Tensor, torch.Tensor]]:
+    ) -> tuple[torch.Tensor, tuple[torch.Tensor, torch.Tensor]]:
         descriptors = descriptors.detach().contiguous()
         projected_descriptors = self.input_projection(descriptors)
         keypoint_encoding_output = self.positional_encoder(keypoints, output_hidden_states=output_hidden_states)
@@ -659,7 +659,7 @@ class LightGlueForKeypointMatching(LightGluePreTrainedModel):
         matches: torch.Tensor,
         matching_scores: torch.Tensor,
         num_keypoints: torch.Tensor,
-    ) -> Tuple[torch.Tensor, torch.Tensor]:
+    ) -> tuple[torch.Tensor, torch.Tensor]:
         # (batch_size, num_keypoints) -> (batch_size // 2, 2, num_keypoints) -> 2 * (batch_size // 2, num_keypoints) to
         # have tensors from
         batch_size, _ = indices.shape
@@ -699,7 +699,7 @@ class LightGlueForKeypointMatching(LightGluePreTrainedModel):
         mask: torch.Tensor = None,
         output_attentions: Optional[bool] = None,
         output_hidden_states: Optional[bool] = None,
-    ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor, Tuple, Tuple]:
+    ) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor, tuple, tuple]:
         all_hidden_states = () if output_hidden_states else None
         all_attentions = () if output_attentions else None
 
@@ -875,7 +875,7 @@ class LightGlueForKeypointMatching(LightGluePreTrainedModel):
         labels: Optional[torch.LongTensor] = None,
         output_attentions: Optional[bool] = None,
         output_hidden_states: Optional[bool] = None,
-    ) -> Union[Tuple, LightGlueKeypointMatchingOutput]:
+    ) -> Union[tuple, LightGlueKeypointMatchingOutput]:
         loss = None
         if labels is not None:
             raise ValueError("LightGlue is not trainable, no labels should be provided.")

src/transformers/models/lightglue/modular_lightglue.py

@@ -12,7 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 from dataclasses import dataclass
-from typing import Callable, Dict, List, Optional, Tuple, Union
+from typing import Callable, Optional, Union
 
 import numpy as np
 import torch
@@ -196,17 +196,17 @@ class LightGlueKeypointMatchingOutput(ModelOutput):
     keypoints: Optional[torch.FloatTensor] = None
     prune: Optional[torch.IntTensor] = None
     mask: 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
 
 
 class LightGlueImageProcessor(SuperGlueImageProcessor):
     def post_process_keypoint_matching(
         self,
         outputs: LightGlueKeypointMatchingOutput,
-        target_sizes: Union[TensorType, List[Tuple]],
+        target_sizes: Union[TensorType, list[tuple]],
         threshold: float = 0.0,
-    ) -> List[Dict[str, torch.Tensor]]:
+    ) -> list[dict[str, torch.Tensor]]:
         return super().post_process_keypoint_matching(outputs, target_sizes, threshold)
 
     def plot_keypoint_matching(self, images: ImageInput, keypoint_matching_output: LightGlueKeypointMatchingOutput):
@@ -263,7 +263,7 @@ class LightGluePositionalEncoder(nn.Module):
 
     def forward(
         self, keypoints: torch.Tensor, output_hidden_states: Optional[bool] = False
-    ) -> Union[Tuple[torch.Tensor], Tuple[torch.Tensor, torch.Tensor]]:
+    ) -> Union[tuple[torch.Tensor], tuple[torch.Tensor, torch.Tensor]]:
         projected_keypoints = self.projector(keypoints)
         embeddings = projected_keypoints.repeat_interleave(2, dim=-1)
         cosines = torch.cos(embeddings)
@@ -277,12 +277,12 @@ class LightGlueAttention(LlamaAttention):
     def forward(
         self,
         hidden_states: torch.Tensor,
-        position_embeddings: Optional[Tuple[torch.Tensor, torch.Tensor]] = None,
+        position_embeddings: Optional[tuple[torch.Tensor, torch.Tensor]] = None,
         attention_mask: Optional[torch.Tensor] = None,
         encoder_hidden_states: Optional[torch.Tensor] = None,
         encoder_attention_mask: Optional[torch.Tensor] = None,
         **kwargs: Unpack[FlashAttentionKwargs],
-    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+    ) -> tuple[torch.Tensor, Optional[torch.Tensor], Optional[tuple[torch.Tensor]]]:
         input_shape = hidden_states.shape[:-1]
         hidden_shape = (*input_shape, -1, self.head_dim)
 
@@ -348,7 +348,7 @@ class LightGlueTransformerLayer(nn.Module):
         attention_mask: torch.Tensor,
         output_hidden_states: Optional[bool] = False,
         output_attentions: Optional[bool] = False,
-    ) -> Tuple[torch.Tensor, Optional[Tuple[torch.Tensor]], Optional[Tuple[torch.Tensor]]]:
+    ) -> tuple[torch.Tensor, Optional[tuple[torch.Tensor]], Optional[tuple[torch.Tensor]]]:
         all_hidden_states = () if output_hidden_states else None
         all_attentions = () if output_attentions else None
 
@@ -509,7 +509,7 @@ class LightGluePreTrainedModel(PreTrainedModel):
             module.weight.data.fill_(1.0)
 
 
-def get_matches_from_scores(scores: torch.Tensor, threshold: float) -> Tuple[torch.Tensor, torch.Tensor]:
+def get_matches_from_scores(scores: torch.Tensor, threshold: float) -> tuple[torch.Tensor, torch.Tensor]:
     """obtain matches from a score matrix [Bx M+1 x N+1]"""
     batch_size, _, _ = scores.shape
     # For each keypoint, get the best match
@@ -622,7 +622,7 @@ class LightGlueForKeypointMatching(LightGluePreTrainedModel):
 
     def _keypoint_processing(
         self, descriptors: torch.Tensor, keypoints: torch.Tensor, output_hidden_states: Optional[bool] = False
-    ) -> Tuple[torch.Tensor, Tuple[torch.Tensor, torch.Tensor]]:
+    ) -> tuple[torch.Tensor, tuple[torch.Tensor, torch.Tensor]]:
         descriptors = descriptors.detach().contiguous()
         projected_descriptors = self.input_projection(descriptors)
         keypoint_encoding_output = self.positional_encoder(keypoints, output_hidden_states=output_hidden_states)
@@ -733,7 +733,7 @@ class LightGlueForKeypointMatching(LightGluePreTrainedModel):
         matches: torch.Tensor,
         matching_scores: torch.Tensor,
         num_keypoints: torch.Tensor,
-    ) -> Tuple[torch.Tensor, torch.Tensor]:
+    ) -> tuple[torch.Tensor, torch.Tensor]:
         # (batch_size, num_keypoints) -> (batch_size // 2, 2, num_keypoints) -> 2 * (batch_size // 2, num_keypoints) to
         # have tensors from
         batch_size, _ = indices.shape
@@ -773,7 +773,7 @@ class LightGlueForKeypointMatching(LightGluePreTrainedModel):
         mask: torch.Tensor = None,
         output_attentions: Optional[bool] = None,
         output_hidden_states: Optional[bool] = None,
-    ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor, Tuple, Tuple]:
+    ) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor, tuple, tuple]:
         all_hidden_states = () if output_hidden_states else None
         all_attentions = () if output_attentions else None
 
@@ -949,7 +949,7 @@ class LightGlueForKeypointMatching(LightGluePreTrainedModel):
         labels: Optional[torch.LongTensor] = None,
         output_attentions: Optional[bool] = None,
         output_hidden_states: Optional[bool] = None,
-    ) -> Union[Tuple, LightGlueKeypointMatchingOutput]:
+    ) -> Union[tuple, LightGlueKeypointMatchingOutput]:
         loss = None
         if labels is not None:
             raise ValueError("LightGlue is not trainable, no labels should be provided.")