apply Black 2024 style in fbcode (4/16)

Summary:
Formats the covered files with pyfmt.

paintitblack

Reviewed By: aleivag

Differential Revision: D54447727

fbshipit-source-id: 8844b1caa08de94d04ac4df3c768dbf8c865fd2f

pytorchvideo/accelerator/deployment/mobile_cpu/transmuter/__init__.py

@@ -5,6 +5,6 @@
 from .transmuter_mobile_cpu import EFFICIENT_BLOCK_TRANSMUTER_MOBILE_CPU
 
 
-EFFICIENT_BLOCK_TRANSMUTER_REGISTRY[
-    "mobile_cpu"
-] = EFFICIENT_BLOCK_TRANSMUTER_MOBILE_CPU
+EFFICIENT_BLOCK_TRANSMUTER_REGISTRY["mobile_cpu"] = (
+    EFFICIENT_BLOCK_TRANSMUTER_MOBILE_CPU
+)

pytorchvideo/data/domsev.py

@@ -197,9 +197,9 @@ def __init__(
             labels_file_path, LabelData, "video_id", list_per_key=True
         )
         # Maps an image frame ID to the singular frame label
-        self._labels_per_frame: Dict[
-            str, int
-        ] = DomsevFrameDataset._assign_labels_to_frames(frames_dict, video_labels)
+        self._labels_per_frame: Dict[str, int] = (
+            DomsevFrameDataset._assign_labels_to_frames(frames_dict, video_labels)
+        )
 
         self._user_transform = transform
         self._transform = self._transform_frame
@@ -380,10 +380,10 @@ def __init__(
             dataset_type,
         )
 
-        self._labels_per_video: Dict[
-            str, List[LabelData]
-        ] = load_dataclass_dict_from_csv(
-            labels_file_path, LabelData, "video_id", list_per_key=True
+        self._labels_per_video: Dict[str, List[LabelData]] = (
+            load_dataclass_dict_from_csv(
+                labels_file_path, LabelData, "video_id", list_per_key=True
+            )
         )
 
         # Sample datapoints

pytorchvideo/layers/squeeze_excitation.py

@@ -16,7 +16,6 @@ def __init__(
         reduction_ratio: Optional[int] = 16,
         reduced_planes: Optional[int] = None,
     ):
-
         """
         Args:
             in_planes (int): input channel dimension.
@@ -74,7 +73,6 @@ def create_audio_2d_squeeze_excitation_block(
     # Activation configs.
     activation: Callable = nn.ReLU,
 ) -> nn.Module:
-
     """
     2-D Residual block with squeeze excitation (SE2D) for 2d. Performs a summation between an
     identity shortcut in branch1 and a main block in branch2. When the input and

pytorchvideo/models/resnet.py

@@ -425,15 +425,17 @@ def create_res_block(
         norm_model = norm(num_features=dim_out, eps=norm_eps, momentum=norm_momentum)
 
     return ResBlock(
-        branch1_conv=conv_skip(
-            dim_in,
-            dim_out,
-            kernel_size=(1, 1, 1),
-            stride=branch1_conv_stride,
-            bias=False,
-        )
-        if (dim_in != dim_out or np.prod(branch1_conv_stride) != 1) or use_shortcut
-        else None,
+        branch1_conv=(
+            conv_skip(
+                dim_in,
+                dim_out,
+                kernel_size=(1, 1, 1),
+                stride=branch1_conv_stride,
+                bias=False,
+            )
+            if (dim_in != dim_out or np.prod(branch1_conv_stride) != 1) or use_shortcut
+            else None
+        ),
         branch1_norm=norm_model,
         branch2=bottleneck(
             dim_in=dim_in,
@@ -795,12 +797,16 @@ def create_resnet(
             conv_b_stride=stage_conv_b_stride,
             conv_b_padding=(
                 stage_conv_b_kernel_size[idx][0] // 2,
-                stage_conv_b_dilation[idx][1]
-                if stage_conv_b_dilation[idx][1] > 1
-                else stage_conv_b_kernel_size[idx][1] // 2,
-                stage_conv_b_dilation[idx][2]
-                if stage_conv_b_dilation[idx][2] > 1
-                else stage_conv_b_kernel_size[idx][2] // 2,
+                (
+                    stage_conv_b_dilation[idx][1]
+                    if stage_conv_b_dilation[idx][1] > 1
+                    else stage_conv_b_kernel_size[idx][1] // 2
+                ),
+                (
+                    stage_conv_b_dilation[idx][2]
+                    if stage_conv_b_dilation[idx][2] > 1
+                    else stage_conv_b_kernel_size[idx][2] // 2
+                ),
             ),
             conv_b_num_groups=stage_conv_b_num_groups[idx],
             conv_b_dilation=stage_conv_b_dilation[idx],

pytorchvideo/models/slowfast.py

@@ -297,12 +297,16 @@ def create_slowfast(
                     conv_b_stride=stage_conv_b_stride,
                     conv_b_padding=(
                         stage_conv_b_kernel_sizes[pathway_idx][idx][0] // 2,
-                        stage_conv_b_dilations[pathway_idx][idx][1]
-                        if stage_conv_b_dilations[pathway_idx][idx][1] > 1
-                        else stage_conv_b_kernel_sizes[pathway_idx][idx][1] // 2,
-                        stage_conv_b_dilations[pathway_idx][idx][2]
-                        if stage_conv_b_dilations[pathway_idx][idx][2] > 1
-                        else stage_conv_b_kernel_sizes[pathway_idx][idx][2] // 2,
+                        (
+                            stage_conv_b_dilations[pathway_idx][idx][1]
+                            if stage_conv_b_dilations[pathway_idx][idx][1] > 1
+                            else stage_conv_b_kernel_sizes[pathway_idx][idx][1] // 2
+                        ),
+                        (
+                            stage_conv_b_dilations[pathway_idx][idx][2]
+                            if stage_conv_b_dilations[pathway_idx][idx][2] > 1
+                            else stage_conv_b_kernel_sizes[pathway_idx][idx][2] // 2
+                        ),
                     ),
                     conv_b_num_groups=stage_conv_b_num_groups[pathway_idx][idx],
                     conv_b_dilation=stage_conv_b_dilations[pathway_idx][idx],

pytorchvideo/models/x3d.py

@@ -294,15 +294,17 @@ def create_x3d_res_block(
         norm_model = norm(num_features=dim_out)
 
     return ResBlock(
-        branch1_conv=nn.Conv3d(
-            dim_in,
-            dim_out,
-            kernel_size=(1, 1, 1),
-            stride=conv_stride,
-            bias=False,
-        )
-        if (dim_in != dim_out or np.prod(conv_stride) > 1) and use_shortcut
-        else None,
+        branch1_conv=(
+            nn.Conv3d(
+                dim_in,
+                dim_out,
+                kernel_size=(1, 1, 1),
+                stride=conv_stride,
+                bias=False,
+            )
+            if (dim_in != dim_out or np.prod(conv_stride) > 1) and use_shortcut
+            else None
+        ),
         branch1_norm=norm_model if dim_in != dim_out and use_shortcut else None,
         branch2=bottleneck(
             dim_in=dim_in,

pytorchvideo/transforms/transforms_factory.py

@@ -222,9 +222,9 @@ def create_video_transform(
                 "scale"
             ]
         if "aspect_ratio" not in random_resized_crop_paras:
-            random_resized_crop_paras[
-                "aspect_ratio"
-            ] = _RANDOM_RESIZED_CROP_DEFAULT_PARAS["aspect_ratio"]
+            random_resized_crop_paras["aspect_ratio"] = (
+                _RANDOM_RESIZED_CROP_DEFAULT_PARAS["aspect_ratio"]
+            )
 
     transform = Compose(
         (

pytorchvideo_trainer/pytorchvideo_trainer/module/ssl_helper.py

@@ -70,7 +70,6 @@ def create_classification_model_from_ssl_checkpoint(
     mlp: Optional[nn.Module] = None,
     detach_backbone: bool = False,
 ) -> nn.Module:
-
     """
     A utlity function for extracting the bacbone from the PyTorch Lightning's
     SSL checkpoints. Used for supervided finetuning the SSL pre-trained models

tests/test_accelerator_models_efficient_x3d.py

@@ -19,7 +19,11 @@ def test_create_x3d(self):
         X3D-M: ("M", 16, 224)
         X3D-L: ("L", 16, 312)
         """
-        for (expansion, input_clip_length, input_crop_size,) in [
+        for (
+            expansion,
+            input_clip_length,
+            input_crop_size,
+        ) in [
             ("XS", 4, 160),
         ]:
             model = create_x3d(expansion=expansion)
@@ -51,7 +55,7 @@ def test_load_hubconf(self):
             os.path.dirname(os.path.realpath(__file__)),
             "..",
         )
-        for (input_clip_length, input_crop_size, model_name) in [
+        for input_clip_length, input_crop_size, model_name in [
             (4, 160, "efficient_x3d_xs"),
             (13, 160, "efficient_x3d_s"),
         ]:

tests/test_data_utils.py

@@ -67,7 +67,10 @@ def _export_video_array(
     ):
         with temp_encoded_video(
             num_frames=num_frames, fps=fps, height=height, width=width
-        ) as (video_file_name, data,), tempfile.TemporaryDirectory(
+        ) as (
+            video_file_name,
+            data,
+        ), tempfile.TemporaryDirectory(
             prefix="video_stop_gap_test"
         ) as tempdir:
             exported_video_path = os.path.join(tempdir, "video.mp4")

tests/test_models_audio_visual_slowfast.py

@@ -19,7 +19,7 @@ def test_create_avslowfast_with_callable(self):
         """
         Test builder `create_audio_visual_slowfast` with callable inputs.
         """
-        for (norm, activation) in itertools.product(
+        for norm, activation in itertools.product(
             (nn.BatchNorm3d, None), (nn.ReLU, nn.Sigmoid, None)
         ):
             input_channel = 3

tests/test_models_head.py

@@ -92,7 +92,7 @@ def test_build_head_with_callable(self):
         """
         Test builder `create_res_basic_head`.
         """
-        for (pool, activation) in itertools.product(
+        for pool, activation in itertools.product(
             (nn.AvgPool3d, nn.MaxPool3d, nn.AdaptiveAvgPool3d, None),
             (nn.ReLU, nn.Softmax, nn.Sigmoid, None),
         ):
@@ -266,7 +266,7 @@ def test_build_complex_head(self):
             )
 
             # Test forwarding.
-            for (input_tensor, bboxes) in TestRoIHeadHelper._get_inputs(
+            for input_tensor, bboxes in TestRoIHeadHelper._get_inputs(
                 input_dim=input_dim
             ):
                 if input_tensor.shape[1] != input_dim:
@@ -298,7 +298,7 @@ def test_build_head_with_callable(self):
             resolution, spatial_scale=spatial_scale, sampling_ratio=sampling_ratio
         )
 
-        for (pool, activation) in itertools.product(
+        for pool, activation in itertools.product(
             (nn.AvgPool3d, nn.MaxPool3d, nn.AdaptiveAvgPool3d, None),
             (nn.ReLU, nn.Softmax, nn.Sigmoid, None),
         ):
@@ -345,7 +345,7 @@ def test_build_head_with_callable(self):
             )  # explicitly use strict mode.
 
             # Test forwarding.
-            for (input_tensor, bboxes) in TestRoIHeadHelper._get_inputs(input_dim=16):
+            for input_tensor, bboxes in TestRoIHeadHelper._get_inputs(input_dim=16):
                 with torch.no_grad():
                     if (
                         input_tensor.shape[1] != 16