linting

stable_baselines3/common/buffers.py

@@ -438,7 +438,7 @@ class NstepReplayBuffer(ReplayBuffer):
         # basically we need to ignore indices that are equal to self.pos
         # because the indice at self.pos is older
         # not_dones that are from other transitions are filtered out.
-        filt = np.hstack([np.ones((len(batch_inds),1)), not_dones[:, 1:]]) * (indices != self.pos)
+        filt = np.hstack([np.ones((len(batch_inds), 1)), not_dones[:, 1:]]) * (indices != self.pos)
         filt = np.multiply.accumulate(filt, 1).reshape(filt.shape)
 
         # weighted rewards

stable_baselines3/common/off_policy_algorithm.py

@@ -21,6 +21,7 @@ from stable_baselines3.common.save_util import save_to_pkl, load_from_pkl
 
 _BufferType = TypeVar("_BufferType", bound=ReplayBuffer)
 
+
 class OffPolicyAlgorithm(BaseAlgorithm, Generic[_BufferType]):
     """
     The base for Off-Policy algorithms (ex: SAC/TD3)

stable_baselines3/dqn/dqn.py

@@ -11,6 +11,7 @@ from stable_baselines3.common.utils import get_linear_fn
 from stable_baselines3.dqn.policies import DQNPolicy
 from stable_baselines3.common.buffers import ReplayBuffer
 
+
 class DQN(OffPolicyAlgorithm):
     """
     Deep Q-Network (DQN)

tests/test_buffers.py

@@ -65,7 +65,7 @@ def test_nsteps():
     # shouldn't be able to get batch with indice 5 because the pointer is at 5
     with pytest.raises(AssertionError):
         buffer._get_samples(np.array([5]))
-    
+
     buffer = NstepReplayBuffer(10, spaces.Discrete(5), spaces.Discrete(5), n_step=5, gamma=0.9)
     buffer.add(0, 1, 10, 1, 1)
     buffer.add(1, 2, 11, 1, 1)