upgrade to black 22.3

Summary: Pull Request resolved: facebookresearch#4268

Reviewed By: zhanghang1989

Differential Revision: D36633678

fbshipit-source-id: 3e25a38eaed1a58c2c03469b36c8cece9de05735

.github/workflows/workflow.yml

@@ -19,7 +19,7 @@ jobs:
         run: |
           python -m pip install --upgrade pip
           python -m pip install flake8==3.8.1 isort==4.3.21
-          python -m pip install black==21.4b2
+          python -m pip install black==22.3.0
           flake8 --version
       - name: Lint
         run: |

detectron2/data/build.py

@@ -552,5 +552,5 @@ def trivial_batch_collator(batch):
 
 
 def worker_init_reset_seed(worker_id):
-    initial_seed = torch.initial_seed() % 2 ** 31
+    initial_seed = torch.initial_seed() % 2**31
     seed_all_rng(initial_seed + worker_id)

detectron2/data/common.py

@@ -141,7 +141,7 @@ def _serialize(data):
             self._addr = np.asarray([len(x) for x in self._lst], dtype=np.int64)
             self._addr = np.cumsum(self._addr)
             self._lst = np.concatenate(self._lst)
-            logger.info("Serialized dataset takes {:.2f} MiB".format(len(self._lst) / 1024 ** 2))
+            logger.info("Serialized dataset takes {:.2f} MiB".format(len(self._lst) / 1024**2))
 
     def __len__(self):
         if self._serialize:

detectron2/engine/defaults.py

@@ -123,7 +123,7 @@ def default_argument_parser(epilog=None):
     # PyTorch still may leave orphan processes in multi-gpu training.
     # Therefore we use a deterministic way to obtain port,
     # so that users are aware of orphan processes by seeing the port occupied.
-    port = 2 ** 15 + 2 ** 14 + hash(os.getuid() if sys.platform != "win32" else 1) % 2 ** 14
+    port = 2**15 + 2**14 + hash(os.getuid() if sys.platform != "win32" else 1) % 2**14
     parser.add_argument(
         "--dist-url",
         default="tcp://127.0.0.1:{}".format(port),

detectron2/evaluation/coco_evaluation.py

@@ -472,14 +472,14 @@ def _evaluate_box_proposals(dataset_predictions, coco_api, thresholds=None, area
         "512-inf": 7,
     }
     area_ranges = [
-        [0 ** 2, 1e5 ** 2],  # all
-        [0 ** 2, 32 ** 2],  # small
-        [32 ** 2, 96 ** 2],  # medium
-        [96 ** 2, 1e5 ** 2],  # large
-        [96 ** 2, 128 ** 2],  # 96-128
-        [128 ** 2, 256 ** 2],  # 128-256
-        [256 ** 2, 512 ** 2],  # 256-512
-        [512 ** 2, 1e5 ** 2],
+        [0**2, 1e5**2],  # all
+        [0**2, 32**2],  # small
+        [32**2, 96**2],  # medium
+        [96**2, 1e5**2],  # large
+        [96**2, 128**2],  # 96-128
+        [128**2, 256**2],  # 128-256
+        [256**2, 512**2],  # 256-512
+        [512**2, 1e5**2],
     ]  # 512-inf
     assert area in areas, "Unknown area range: {}".format(area)
     area_range = area_ranges[areas[area]]

detectron2/evaluation/lvis_evaluation.py

@@ -238,14 +238,14 @@ def _evaluate_box_proposals(dataset_predictions, lvis_api, thresholds=None, area
         "512-inf": 7,
     }
     area_ranges = [
-        [0 ** 2, 1e5 ** 2],  # all
-        [0 ** 2, 32 ** 2],  # small
-        [32 ** 2, 96 ** 2],  # medium
-        [96 ** 2, 1e5 ** 2],  # large
-        [96 ** 2, 128 ** 2],  # 96-128
-        [128 ** 2, 256 ** 2],  # 128-256
-        [256 ** 2, 512 ** 2],  # 256-512
-        [512 ** 2, 1e5 ** 2],
+        [0**2, 1e5**2],  # all
+        [0**2, 32**2],  # small
+        [32**2, 96**2],  # medium
+        [96**2, 1e5**2],  # large
+        [96**2, 128**2],  # 96-128
+        [128**2, 256**2],  # 128-256
+        [256**2, 512**2],  # 256-512
+        [512**2, 1e5**2],
     ]  # 512-inf
     assert area in areas, "Unknown area range: {}".format(area)
     area_range = area_ranges[areas[area]]

detectron2/export/c10.py

@@ -2,7 +2,6 @@
 
 import math
 from typing import Dict
-
 import torch
 import torch.nn.functional as F
 
@@ -164,12 +163,12 @@ def _set_tensor_mode(self, v):
 
 
 class Caffe2RPN(Caffe2Compatible, rpn.RPN):
-
     @classmethod
     def from_config(cls, cfg, input_shape: Dict[str, ShapeSpec]):
         ret = super(Caffe2Compatible, cls).from_config(cfg, input_shape)
-        assert tuple(cfg.MODEL.RPN.BBOX_REG_WEIGHTS) == (1., 1., 1., 1.) or \
-            tuple(cfg.MODEL.RPN.BBOX_REG_WEIGHTS) == (1., 1., 1., 1., 1.)
+        assert tuple(cfg.MODEL.RPN.BBOX_REG_WEIGHTS) == (1.0, 1.0, 1.0, 1.0) or tuple(
+            cfg.MODEL.RPN.BBOX_REG_WEIGHTS
+        ) == (1.0, 1.0, 1.0, 1.0, 1.0)
         return ret
 
     def _generate_proposals(

detectron2/layers/losses.py

@@ -119,7 +119,7 @@ def ciou_loss(
     h_pred = y2 - y1
     w_gt = x2g - x1g
     h_gt = y2g - y1g
-    v = (4 / (math.pi ** 2)) * torch.pow((torch.atan(w_gt / h_gt) - torch.atan(w_pred / h_pred)), 2)
+    v = (4 / (math.pi**2)) * torch.pow((torch.atan(w_gt / h_gt) - torch.atan(w_pred / h_pred)), 2)
     with torch.no_grad():
         alpha = v / (1 - iou + v + eps)
 

detectron2/layers/mask_ops.py

@@ -11,7 +11,7 @@
 BYTES_PER_FLOAT = 4
 # TODO: This memory limit may be too much or too little. It would be better to
 # determine it based on available resources.
-GPU_MEM_LIMIT = 1024 ** 3  # 1 GB memory limit
+GPU_MEM_LIMIT = 1024**3  # 1 GB memory limit
 
 
 def _do_paste_mask(masks, boxes, img_h: int, img_w: int, skip_empty: bool = True):

detectron2/modeling/anchor_generator.py

@@ -202,7 +202,7 @@ def generate_cell_anchors(self, sizes=(32, 64, 128, 256, 512), aspect_ratios=(0.
 
         anchors = []
         for size in sizes:
-            area = size ** 2.0
+            area = size**2.0
             for aspect_ratio in aspect_ratios:
                 # s * s = w * h
                 # a = h / w
@@ -349,7 +349,7 @@ def generate_cell_anchors(
         """
         anchors = []
         for size in sizes:
-            area = size ** 2.0
+            area = size**2.0
             for aspect_ratio in aspect_ratios:
                 # s * s = w * h
                 # a = h / w

detectron2/solver/build.py

@@ -267,7 +267,7 @@ def build_lr_scheduler(
                 "These values will be ignored."
             )
         sched = MultiStepParamScheduler(
-            values=[cfg.SOLVER.GAMMA ** k for k in range(len(steps) + 1)],
+            values=[cfg.SOLVER.GAMMA**k for k in range(len(steps) + 1)],
             milestones=steps,
             num_updates=cfg.SOLVER.MAX_ITER,
         )

detectron2/utils/comm.py

@@ -162,7 +162,7 @@ def shared_random_seed():
 
     All workers must call this function, otherwise it will deadlock.
     """
-    ints = np.random.randint(2 ** 31)
+    ints = np.random.randint(2**31)
     all_ints = all_gather(ints)
     return all_ints[0]
 

detectron2/utils/events.py

@@ -367,7 +367,7 @@ def put_histogram(self, hist_name, hist_tensor, bins=1000):
             max=ht_max,
             num=len(hist_tensor),
             sum=float(hist_tensor.sum()),
-            sum_squares=float(torch.sum(hist_tensor ** 2)),
+            sum_squares=float(torch.sum(hist_tensor**2)),
             bucket_limits=hist_edges[1:].tolist(),
             bucket_counts=hist_counts.tolist(),
             global_step=self._iter,

dev/linter.sh

@@ -5,9 +5,9 @@
 cd "$(dirname "${BASH_SOURCE[0]}")/.."
 
 {
-  black --version | grep -E "21\." > /dev/null
+  black --version | grep -E "22\." > /dev/null
 } || {
-  echo "Linter requires 'black==21.*' !"
+  echo "Linter requires 'black==22.*' !"
   exit 1
 }
 

projects/DensePose/densepose/evaluation/densepose_coco_evaluation.py

@@ -527,7 +527,7 @@ def computeOks(self, imgId, catId):
                     z = np.zeros(k)
                     dx = np.max((z, x0 - xd), axis=0) + np.max((z, xd - x1), axis=0)
                     dy = np.max((z, y0 - yd), axis=0) + np.max((z, yd - y1), axis=0)
-                e = (dx ** 2 + dy ** 2) / vars / (gt["area"] + np.spacing(1)) / 2
+                e = (dx**2 + dy**2) / vars / (gt["area"] + np.spacing(1)) / 2
                 if k1 > 0:
                     e = e[vg > 0]
                 ious[i, j] = np.sum(np.exp(-e)) / e.shape[0]
@@ -767,7 +767,7 @@ def computeOgps(self, imgId, catId):
                         )
                         # Compute gps
                         ogps_values = np.exp(
-                            -(dists_between_matches ** 2) / (2 * (dist_norm_coeffs ** 2))
+                            -(dists_between_matches**2) / (2 * (dist_norm_coeffs**2))
                         )
                         #
                         ogps = np.mean(ogps_values) if len(ogps_values) > 0 else 0.0
@@ -1265,10 +1265,10 @@ def setDetParams(self):
         self.recThrs = np.linspace(0.0, 1.00, int(np.round((1.00 - 0.0) / 0.01)) + 1, endpoint=True)
         self.maxDets = [1, 10, 100]
         self.areaRng = [
-            [0 ** 2, 1e5 ** 2],
-            [0 ** 2, 32 ** 2],
-            [32 ** 2, 96 ** 2],
-            [96 ** 2, 1e5 ** 2],
+            [0**2, 1e5**2],
+            [0**2, 32**2],
+            [32**2, 96**2],
+            [96**2, 1e5**2],
         ]
         self.areaRngLbl = ["all", "small", "medium", "large"]
         self.useCats = 1
@@ -1280,7 +1280,7 @@ def setKpParams(self):
         self.iouThrs = np.linspace(0.5, 0.95, np.round((0.95 - 0.5) / 0.05) + 1, endpoint=True)
         self.recThrs = np.linspace(0.0, 1.00, np.round((1.00 - 0.0) / 0.01) + 1, endpoint=True)
         self.maxDets = [20]
-        self.areaRng = [[0 ** 2, 1e5 ** 2], [32 ** 2, 96 ** 2], [96 ** 2, 1e5 ** 2]]
+        self.areaRng = [[0**2, 1e5**2], [32**2, 96**2], [96**2, 1e5**2]]
         self.areaRngLbl = ["all", "medium", "large"]
         self.useCats = 1
 
@@ -1290,7 +1290,7 @@ def setUvParams(self):
         self.iouThrs = np.linspace(0.5, 0.95, int(np.round((0.95 - 0.5) / 0.05)) + 1, endpoint=True)
         self.recThrs = np.linspace(0.0, 1.00, int(np.round((1.00 - 0.0) / 0.01)) + 1, endpoint=True)
         self.maxDets = [20]
-        self.areaRng = [[0 ** 2, 1e5 ** 2], [32 ** 2, 96 ** 2], [96 ** 2, 1e5 ** 2]]
+        self.areaRng = [[0**2, 1e5**2], [32**2, 96**2], [96**2, 1e5**2]]
         self.areaRngLbl = ["all", "medium", "large"]
         self.useCats = 1
 

projects/DensePose/densepose/evaluation/mesh_alignment_evaluator.py

@@ -49,7 +49,7 @@ def evaluate(self):
                     [keyvertices_2[name] for name in keyvertex_names_1],
                 ]
                 Current_Mean_Distances = 0.255
-                gps = (-(geodists ** 2) / (2 * (Current_Mean_Distances ** 2))).exp()
+                gps = (-(geodists**2) / (2 * (Current_Mean_Distances**2))).exp()
                 avg_errors.append(geodists.mean().item())
                 avg_gps.append(gps.mean().item())
 

projects/DensePose/densepose/modeling/hrfpn.py

@@ -90,7 +90,7 @@ def __init__(
                         in_channels=in_channels[i],
                         out_channels=in_channels[i],
                         kernel_size=4,
-                        stride=2 ** i,
+                        stride=2**i,
                         padding=0,
                         output_padding=0,
                         bias=False,
@@ -105,7 +105,7 @@ def __init__(
         for i in range(self.n_out_features):
             self.reduction_pooling_conv.append(
                 nn.Sequential(
-                    nn.Conv2d(sum(in_channels), out_channels, kernel_size=2 ** i, stride=2 ** i),
+                    nn.Conv2d(sum(in_channels), out_channels, kernel_size=2**i, stride=2**i),
                     nn.BatchNorm2d(out_channels, momentum=0.1),
                     nn.ReLU(inplace=True),
                 )
@@ -148,7 +148,7 @@ def forward(self, inputs):
             outs.append(self.reduction_pooling_conv[i](out))
         for i in range(len(outs)):  # Make shapes consistent
             outs[-1 - i] = outs[-1 - i][
-                :, :, : outs[-1].shape[2] * 2 ** i, : outs[-1].shape[3] * 2 ** i
+                :, :, : outs[-1].shape[2] * 2**i, : outs[-1].shape[3] * 2**i
             ]
         outputs = []
         for i in range(len(outs)):

projects/DensePose/densepose/modeling/losses/chart_with_confidences.py

@@ -188,17 +188,17 @@ def forward(
         # compute $\sigma_i^2$
         sigma2 = F.softplus(sigma_u) + self.sigma_lower_bound
         # compute \|r_i\|^2
-        r_sqnorm2 = kappa_u_est ** 2 + kappa_v_est ** 2
+        r_sqnorm2 = kappa_u_est**2 + kappa_v_est**2
         delta_u = u - target_u
         delta_v = v - target_v
         # compute \|delta_i\|^2
-        delta_sqnorm = delta_u ** 2 + delta_v ** 2
+        delta_sqnorm = delta_u**2 + delta_v**2
         delta_u_r_u = delta_u * kappa_u_est
         delta_v_r_v = delta_v * kappa_v_est
         # compute the scalar product <delta_i, r_i>
         delta_r = delta_u_r_u + delta_v_r_v
         # compute squared scalar product <delta_i, r_i>^2
-        delta_r_sqnorm = delta_r ** 2
+        delta_r_sqnorm = delta_r**2
         denom2 = sigma2 * (sigma2 + r_sqnorm2)
         loss = 0.5 * (
             self.log2pi + torch.log(denom2) + delta_sqnorm / sigma2 - delta_r_sqnorm / denom2

projects/DensePose/densepose/modeling/test_time_augmentation.py

@@ -193,11 +193,11 @@ def rotate_box_inverse(rot_tfm, rotated_box):
     invrot_box = rot_tfm.inverse().apply_box(rotated_box)
     h, w = rotated_box[:, 3] - rotated_box[:, 1], rotated_box[:, 2] - rotated_box[:, 0]
     ih, iw = invrot_box[:, 3] - invrot_box[:, 1], invrot_box[:, 2] - invrot_box[:, 0]
-    assert 2 * rot_tfm.abs_sin ** 2 != 1, "45 degrees angle can't be inverted"
+    assert 2 * rot_tfm.abs_sin**2 != 1, "45 degrees angle can't be inverted"
     # 2. Inverse the corresponding computation in the rotation transform
     # to get the original height/width of the rotated boxes
-    orig_h = (h * rot_tfm.abs_cos - w * rot_tfm.abs_sin) / (1 - 2 * rot_tfm.abs_sin ** 2)
-    orig_w = (w * rot_tfm.abs_cos - h * rot_tfm.abs_sin) / (1 - 2 * rot_tfm.abs_sin ** 2)
+    orig_h = (h * rot_tfm.abs_cos - w * rot_tfm.abs_sin) / (1 - 2 * rot_tfm.abs_sin**2)
+    orig_w = (w * rot_tfm.abs_cos - h * rot_tfm.abs_sin) / (1 - 2 * rot_tfm.abs_sin**2)
     # 3. Resize the inverse-rotated bboxes to their original size
     invrot_box[:, 0] += (iw - orig_w) / 2
     invrot_box[:, 1] += (ih - orig_h) / 2

projects/DensePose/densepose/vis/densepose_outputs_vertex.py

@@ -33,7 +33,7 @@ def get_xyz_vertex_embedding(mesh_name: str, device: torch.device):
         embed_map = mesh.vertices.sum(dim=1)
         embed_map -= embed_map.min()
         embed_map /= embed_map.max()
-        embed_map = embed_map ** 2
+        embed_map = embed_map**2
     return embed_map
 
 

projects/Panoptic-DeepLab/panoptic_deeplab/target_generator.py

@@ -47,7 +47,7 @@ def __init__(
         x = np.arange(0, size, 1, float)
         y = x[:, np.newaxis]
         x0, y0 = 3 * sigma + 1, 3 * sigma + 1
-        self.g = np.exp(-((x - x0) ** 2 + (y - y0) ** 2) / (2 * sigma ** 2))
+        self.g = np.exp(-((x - x0) ** 2 + (y - y0) ** 2) / (2 * sigma**2))
 
     def __call__(self, panoptic, segments_info):
         """Generates the training target.

projects/PointRend/point_rend/mask_head.py

@@ -210,7 +210,7 @@ def _init_point_head(self, cfg, input_shape):
         # An optimization to skip unused subdivision steps: if after subdivision, all pixels on
         # the mask will be selected and recomputed anyway, we should just double our init_resolution
         while (
-            4 * self.mask_point_subdivision_init_resolution ** 2
+            4 * self.mask_point_subdivision_init_resolution**2
             <= self.mask_point_subdivision_num_points
         ):
             self.mask_point_subdivision_init_resolution *= 2
@@ -404,7 +404,7 @@ def forward(self, features, instances):
         if self.training:
             proposal_boxes = [x.proposal_boxes for x in instances]
             parameters = self.parameter_head(self._roi_pooler(features, proposal_boxes))
-            losses = {"loss_l2": self.regularizer * (parameters ** 2).mean()}
+            losses = {"loss_l2": self.regularizer * (parameters**2).mean()}
 
             point_coords, point_labels = self._uniform_sample_train_points(instances)
             point_fine_grained_features = self._point_pooler(features, proposal_boxes, point_coords)

projects/TensorMask/tensormask/arch.py

@@ -468,7 +468,7 @@ def losses(
         if self.mask_on:
             loss_mask = 0
             for lvl in range(self.num_levels):
-                cur_level_factor = 2 ** lvl if self.bipyramid_on else 1
+                cur_level_factor = 2**lvl if self.bipyramid_on else 1
                 for anc in range(self.num_anchors):
                     cur_gt_mask_inds = gt_mask_inds[lvl][anc]
                     if cur_gt_mask_inds is None:
@@ -477,7 +477,7 @@ def losses(
                         cur_mask_size = self.mask_sizes[anc] * cur_level_factor
                         # TODO maybe there are numerical issues when mask sizes are large
                         cur_size_divider = torch.tensor(
-                            self.mask_loss_weight / (cur_mask_size ** 2),
+                            self.mask_loss_weight / (cur_mask_size**2),
                             dtype=torch.float32,
                             device=self.device,
                         )
@@ -591,7 +591,7 @@ def get_ground_truth(self, anchors, unit_lengths, indexes, targets):
                     for lvl in range(self.num_levels):
                         ids_lvl = matched_indexes[:, 0] == lvl
                         if torch.any(ids_lvl):
-                            cur_level_factor = 2 ** lvl if self.bipyramid_on else 1
+                            cur_level_factor = 2**lvl if self.bipyramid_on else 1
                             for anc in range(self.num_anchors):
                                 ids_lvl_anchor = ids_lvl & (matched_indexes[:, 4] == anc)
                                 if torch.any(ids_lvl_anchor):
@@ -734,7 +734,7 @@ def inference_single_image(
             result_anchors = top_anchors[keep]
             # Get masks and do sigmoid
             for lvl, _, h, w, anc in result_indexes.tolist():
-                cur_size = self.mask_sizes[anc] * (2 ** lvl if self.bipyramid_on else 1)
+                cur_size = self.mask_sizes[anc] * (2**lvl if self.bipyramid_on else 1)
                 result_masks.append(
                     torch.sigmoid(pred_masks[lvl][anc][:, h, w].view(1, cur_size, cur_size))
                 )
@@ -831,7 +831,7 @@ def __init__(self, cfg, num_levels, num_anchors, mask_sizes, input_shape: List[S
                 if self.bipyramid_on:
                     for lvl in range(num_levels):
                         cur_mask_module = "align2nat_%02d" % lvl
-                        lambda_val = 2 ** lvl
+                        lambda_val = 2**lvl
                         setattr(self, cur_mask_module, SwapAlign2Nat(lambda_val))
                     # Also the fusing layer, stay at the same channel size
                     mask_fuse = [
@@ -885,7 +885,7 @@ def forward(self, features):
                 H, W = mask_feat_high_res.shape[-2:]
                 mask_feats_up = []
                 for lvl, mask_feat in enumerate(mask_feats):
-                    lambda_val = 2.0 ** lvl
+                    lambda_val = 2.0**lvl
                     mask_feat_up = mask_feat
                     if lvl > 0:
                         mask_feat_up = F.interpolate(