Black reformatting

Author: bmind7

ml-agents/mlagents/trainers/torch_entities/components/reward_providers/gail_reward_provider.py

@@ -226,7 +226,9 @@ def compute_gradient_magnitude(
         if self._settings.use_actions:
             policy_action = self.get_action_input(policy_batch)
             expert_action = self.get_action_input(expert_batch)
-            action_epsilon = torch.rand(policy_action.shape, device=policy_action.device)
+            action_epsilon = torch.rand(
+                policy_action.shape, device=policy_action.device
+            )
             policy_dones = torch.as_tensor(
                 policy_batch[BufferKey.DONE], dtype=torch.float, device=default_device()
             ).unsqueeze(1)

ml-agents/mlagents/trainers/torch_entities/networks.py

@@ -86,10 +86,10 @@ def total_goal_enc_size(self) -> int:
     def update_normalization(self, buffer: AgentBuffer) -> None:
         obs = ObsUtil.from_buffer(buffer, len(self.processors))
         for vec_input, enc in zip(obs, self.processors):
-                if isinstance(enc, VectorInput):
-                    enc.update_normalization(
-                        torch.as_tensor(vec_input.to_ndarray(), device=default_device())
-                    )
+            if isinstance(enc, VectorInput):
+                enc.update_normalization(
+                    torch.as_tensor(vec_input.to_ndarray(), device=default_device())
+                )
 
     def copy_normalization(self, other_encoder: "ObservationEncoder") -> None:
         if self.normalize:

ml-agents/mlagents/trainers/torch_entities/utils.py

@@ -234,9 +234,7 @@ def list_to_tensor(
         calling as_tensor on the list directly.
         """
         device = default_device()
-        return torch.as_tensor(
-            np.asanyarray(ndarray_list), dtype=dtype, device=device
-        )
+        return torch.as_tensor(np.asanyarray(ndarray_list), dtype=dtype, device=device)
 
     @staticmethod
     def list_to_tensor_list(