Fix default arguments + add bugbear (DLR-RM#363)

* Fix potential bug + add bug bear

* Remove unused variables

* Minor: version bump

Author: araffin

docs/misc/changelog.rst

@@ -4,7 +4,7 @@ Changelog
 ==========
 
 
-Release 1.1.0a0 (WIP)
+Release 1.1.0a1 (WIP)
 ---------------------------
 
 Breaking Changes:
@@ -15,13 +15,14 @@ New Features:
 
 Bug Fixes:
 ^^^^^^^^^^
-- Fix GAE computation for on-policy algorithms (off-by one for the last value) (thanks @Wovchena)
+- Fixed potential issue when calling off-policy algorithms with default arguments multiple times (the size of the replay buffer would be the same)
 
 Deprecations:
 ^^^^^^^^^^^^^
 
 Others:
 ^^^^^^^
+- Added ``flake8-bugbear`` to tests dependencies to find likely bugs
 
 Documentation:
 ^^^^^^^^^^^^^^

setup.py

@@ -20,8 +20,8 @@
 Repository:
 https://github.com/DLR-RM/stable-baselines3
 
-Medium article:
-https://medium.com/@araffin/df87c4b2fc82
+Blog post:
+https://araffin.github.io/post/sb3/
 
 Documentation:
 https://stable-baselines3.readthedocs.io/en/master/
@@ -94,6 +94,8 @@
             "pytype",
             # Lint code
             "flake8>=3.8",
+            # Find likely bugs
+            "flake8-bugbear",
             # Sort imports
             "isort>=5.0",
             # Reformat

stable_baselines3/common/off_policy_algorithm.py

@@ -76,7 +76,7 @@ def __init__(
         env: Union[GymEnv, str],
         policy_base: Type[BasePolicy],
         learning_rate: Union[float, Schedule],
-        buffer_size: int = int(1e6),
+        buffer_size: int = 1000000,
         learning_starts: int = 100,
         batch_size: int = 256,
         tau: float = 0.005,

stable_baselines3/common/results_plotter.py

@@ -84,7 +84,7 @@ def plot_curves(
     plt.figure(title, figsize=figsize)
     max_x = max(xy[0][-1] for xy in xy_list)
     min_x = 0
-    for (i, (x, y)) in enumerate(xy_list):
+    for (_, (x, y)) in enumerate(xy_list):
         plt.scatter(x, y, s=2)
         # Do not plot the smoothed curve at all if the timeseries is shorter than window size.
         if x.shape[0] >= EPISODES_WINDOW:

stable_baselines3/common/torch_layers.py

@@ -170,7 +170,7 @@ def __init__(
         last_layer_dim_shared = feature_dim
 
         # Iterate through the shared layers and build the shared parts of the network
-        for idx, layer in enumerate(net_arch):
+        for layer in net_arch:
             if isinstance(layer, int):  # Check that this is a shared layer
                 layer_size = layer
                 # TODO: give layer a meaningful name
@@ -192,7 +192,7 @@ def __init__(
         last_layer_dim_vf = last_layer_dim_shared
 
         # Build the non-shared part of the network
-        for idx, (pi_layer_size, vf_layer_size) in enumerate(zip_longest(policy_only_layers, value_only_layers)):
+        for pi_layer_size, vf_layer_size in zip_longest(policy_only_layers, value_only_layers):
             if pi_layer_size is not None:
                 assert isinstance(pi_layer_size, int), "Error: net_arch[-1]['pi'] must only contain integers."
                 policy_net.append(nn.Linear(last_layer_dim_pi, pi_layer_size))

stable_baselines3/ddpg/ddpg.py

@@ -54,7 +54,7 @@ def __init__(
         policy: Union[str, Type[TD3Policy]],
         env: Union[GymEnv, str],
         learning_rate: Union[float, Schedule] = 1e-3,
-        buffer_size: int = int(1e6),
+        buffer_size: int = 1000000,
         learning_starts: int = 100,
         batch_size: int = 100,
         tau: float = 0.005,

stable_baselines3/dqn/dqn.py

@@ -147,7 +147,7 @@ def train(self, gradient_steps: int, batch_size: int = 100) -> None:
         self._update_learning_rate(self.policy.optimizer)
 
         losses = []
-        for gradient_step in range(gradient_steps):
+        for _ in range(gradient_steps):
             # Sample replay buffer
             replay_data = self.replay_buffer.sample(batch_size, env=self._vec_normalize_env)
 

stable_baselines3/sac/sac.py

@@ -74,7 +74,7 @@ def __init__(
         policy: Union[str, Type[SACPolicy]],
         env: Union[GymEnv, str],
         learning_rate: Union[float, Schedule] = 3e-4,
-        buffer_size: int = int(1e6),
+        buffer_size: int = 1000000,
         learning_starts: int = 100,
         batch_size: int = 256,
         tau: float = 0.005,

stable_baselines3/td3/td3.py

@@ -62,7 +62,7 @@ def __init__(
         policy: Union[str, Type[TD3Policy]],
         env: Union[GymEnv, str],
         learning_rate: Union[float, Schedule] = 1e-3,
-        buffer_size: int = int(1e6),
+        buffer_size: int = 1000000,  # 1e6
         learning_starts: int = 100,
         batch_size: int = 100,
         tau: float = 0.005,
@@ -131,7 +131,7 @@ def train(self, gradient_steps: int, batch_size: int = 100) -> None:
 
         actor_losses, critic_losses = [], []
 
-        for gradient_step in range(gradient_steps):
+        for _ in range(gradient_steps):
 
             self._n_updates += 1
             # Sample replay buffer

stable_baselines3/version.txt

@@ -1 +1 @@
-1.1.0a0
+1.1.0a1