diff --git a/examples/dcgan.py b/examples/dcgan.py
index c1065f12..585e145a 100644
--- a/examples/dcgan.py
+++ b/examples/dcgan.py
@@ -201,7 +201,7 @@ def __init__(self, ngpu):
             nn.ReLU(True),
             # state size. (ngf) x 32 x 32
             nn.ConvTranspose2d(ngf, nc, 4, 2, 1, bias=False),
-            nn.Tanh()
+            nn.Tanh(),
             # state size. (nc) x 64 x 64
         )
 
diff --git a/opacus/accountants/analysis/prv/domain.py b/opacus/accountants/analysis/prv/domain.py
index 5e86502d..220dbbd9 100644
--- a/opacus/accountants/analysis/prv/domain.py
+++ b/opacus/accountants/analysis/prv/domain.py
@@ -12,6 +12,7 @@ class Domain:
     Stores relevant information about the domain on which PRVs are discretized, and
     includes a few convenience methods for manipulating it.
     """
+
     t_min: float
     t_max: float
     size: int
diff --git a/opacus/layers/dp_rnn.py b/opacus/layers/dp_rnn.py
index 5f5c6819..d1d5b61f 100644
--- a/opacus/layers/dp_rnn.py
+++ b/opacus/layers/dp_rnn.py
@@ -405,9 +405,10 @@ def forward(
             for direction, (cell, h0, c0) in directions:
                 # apply single direction layer (with dropout)
                 out_layer, h, c = self.forward_layer(
-                    x
-                    if layer == 0
-                    else output,  # [T, B, D/H/2H] / tuple T x [B, D/H/2H]
+                    (
+                        x if layer == 0 else output
+                        # [T, B, D/H/2H] / tuple T x [B, D/H/2H]
+                    ),
                     h0,  # [B, H]
                     c0,
                     batch_sizes,
diff --git a/opacus/privacy_engine.py b/opacus/privacy_engine.py
index 0ca6811b..555769d0 100644
--- a/opacus/privacy_engine.py
+++ b/opacus/privacy_engine.py
@@ -536,9 +536,9 @@ def save_checkpoint(
         if noise_scheduler is not None:
             checkpoint_dict["noise_scheduler_state_dict"] = noise_scheduler.state_dict()
         if grad_clip_scheduler is not None:
-            checkpoint_dict[
-                "grad_clip_scheduler_state_dict"
-            ] = grad_clip_scheduler.state_dict()
+            checkpoint_dict["grad_clip_scheduler_state_dict"] = (
+                grad_clip_scheduler.state_dict()
+            )
 
         torch.save(checkpoint_dict, path, **(torch_save_kwargs or {}))
 
diff --git a/opacus/tests/grad_samples/conv2d_test.py b/opacus/tests/grad_samples/conv2d_test.py
index b0deae3f..d1178506 100644
--- a/opacus/tests/grad_samples/conv2d_test.py
+++ b/opacus/tests/grad_samples/conv2d_test.py
@@ -90,9 +90,9 @@ def test_conv2d(
             # Test 'convolution as a backward' GSM
             # 'convolution as a backward' doesn't support padding=same
             conv2d_gsm = GradSampleModule.GRAD_SAMPLERS[nn.Conv2d]
-            GradSampleModule.GRAD_SAMPLERS[
-                nn.Conv2d
-            ] = convolution2d_backward_as_a_convolution
+            GradSampleModule.GRAD_SAMPLERS[nn.Conv2d] = (
+                convolution2d_backward_as_a_convolution
+            )
             self.run_test(
                 x,
                 conv,
diff --git a/opacus/tests/privacy_engine_test.py b/opacus/tests/privacy_engine_test.py
index 022acfa3..f2ed1a32 100644
--- a/opacus/tests/privacy_engine_test.py
+++ b/opacus/tests/privacy_engine_test.py
@@ -87,9 +87,11 @@ def _init_vanilla_training(
     ):
         model = self._init_model()
         optimizer = torch.optim.SGD(
-            model.parameters()
-            if not opt_exclude_frozen
-            else [p for p in model.parameters() if p.requires_grad],
+            (
+                model.parameters()
+                if not opt_exclude_frozen
+                else [p for p in model.parameters() if p.requires_grad]
+            ),
             lr=self.LR,
             momentum=0,
         )
@@ -112,9 +114,11 @@ def _init_private_training(
         model = self._init_model()
         model = PrivacyEngine.get_compatible_module(model)
         optimizer = torch.optim.SGD(
-            model.parameters()
-            if not opt_exclude_frozen
-            else [p for p in model.parameters() if p.requires_grad],
+            (
+                model.parameters()
+                if not opt_exclude_frozen
+                else [p for p in model.parameters() if p.requires_grad]
+            ),
             lr=self.LR,
             momentum=0,
         )