Fix getattr

bsuite/baselines/jax/actor_critic/agent.py

@@ -56,7 +56,7 @@ def __init__(
     # Define loss function.
     def loss(trajectory: sequence.Trajectory) -> jnp.ndarray:
       """"Actor-critic loss."""
-      logits, values = network(trajectory.observations)
+      logits, values = network(trajectory.observations)  # pytype: disable=wrong-arg-types  # jax-ndarray
       td_errors = rlax.td_lambda(
           v_tm1=values[:-1],
           r_t=trajectory.rewards,

bsuite/baselines/utils/sequence_test.py

@@ -31,8 +31,8 @@ def test_buffer(self):
     max_sequence_length = 10
     obs_shape = (3, 3)
     buffer = sequence.Buffer(
-        obs_spec=specs.Array(obs_shape, dtype=np.float),
-        action_spec=specs.Array((), dtype=np.int),
+        obs_spec=specs.Array(obs_shape, dtype=float),
+        action_spec=specs.Array((), dtype=int),
         max_sequence_length=max_sequence_length)
     dummy_step = dm_env.transition(observation=np.zeros(obs_shape), reward=0.)
 

bsuite/environments/cartpole.py

@@ -159,10 +159,10 @@ def _reset(self) -> dm_env.TimeStep:
     raise NotImplementedError('This environment implements its own auto-reset.')
 
   def action_spec(self):
-    return specs.DiscreteArray(dtype=np.int, num_values=3, name='action')
+    return specs.DiscreteArray(dtype=int, num_values=3, name='action')
 
   def observation_spec(self):
-    return specs.Array(shape=(1, 6), dtype=np.float32, name='state')
+    return specs.Array(shape=(1, 6), dtype=np.float32, name='observation')
 
   @property
   def observation(self) -> np.ndarray:

bsuite/environments/catch.py

@@ -99,12 +99,12 @@ def _step(self, action: int) -> dm_env.TimeStep:
   def observation_spec(self) -> specs.BoundedArray:
     """Returns the observation spec."""
     return specs.BoundedArray(shape=self._board.shape, dtype=self._board.dtype,
-                              name="board", minimum=0, maximum=1)
+                              name="observation", minimum=0, maximum=1)
 
   def action_spec(self) -> specs.DiscreteArray:
     """Returns the action spec."""
     return specs.DiscreteArray(
-        dtype=np.int, num_values=len(_ACTIONS), name="action")
+        dtype=int, num_values=len(_ACTIONS), name="action")
 
   def _observation(self) -> np.ndarray:
     self._board.fill(0.)

bsuite/experiments/cartpole_swingup/cartpole_swingup.py

@@ -129,7 +129,7 @@ def _reset(self) -> dm_env.TimeStep:
     raise NotImplementedError('This environment implements its own auto-reset.')
 
   def action_spec(self):
-    return specs.DiscreteArray(dtype=np.int, num_values=3, name='action')
+    return specs.DiscreteArray(dtype=int, num_values=3, name='action')
 
   def observation_spec(self):
     return specs.Array(shape=(1, 8), dtype=np.float32, name='state')

bsuite/experiments/deep_sea/analysis.py

@@ -37,7 +37,7 @@ def _check_data(df: pd.DataFrame) -> None:
 def find_solution(df_in: pd.DataFrame,
                   sweep_vars: Optional[Sequence[str]] = None,
                   merge: bool = True,
-                  thresh: float = 0.8,
+                  thresh: float = 0.9,
                   num_episodes: int = NUM_EPISODES) -> pd.DataFrame:
   """Find first episode that gets below thresh regret by sweep_vars."""
   # Check data has the necessary columns for deep sea

bsuite/logging/logging_utils.py

@@ -15,7 +15,7 @@
 # ============================================================================
 """Read functionality for local csv-based experiments."""
 
-import collections
+from collections import abc
 import copy
 from typing import Any, Callable, List, Mapping, Sequence, Tuple, Union
 
@@ -72,7 +72,7 @@ def load_multiple_runs(
   # Convert any inputs to dictionary format.
   if isinstance(path_collection, six.string_types):
     path_collection = {path_collection: path_collection}
-  if not isinstance(path_collection, collections.Mapping):
+  if not isinstance(path_collection, abc.Mapping):
     path_collection = {path: path for path in path_collection}
 
   # Loop through multiple bsuite runs, and apply single_load_fn to each.

bsuite/utils/gym_wrapper.py

@@ -97,8 +97,9 @@ def reward_range(self) -> Tuple[float, float]:
 
   def __getattr__(self, attr):
     """Delegate attribute access to underlying environment."""
-    return getattr(self._env, attr)
-
+    if "_env" in self.__dict__:
+      return getattr(self._env, attr)
+    return super().__getattribute__(attr)
 
 def space2spec(space: gym.Space, name: Optional[str] = None):
   """Converts an OpenAI Gym space to a dm_env spec or nested structure of specs.

bsuite/utils/wrappers.py

@@ -134,8 +134,9 @@ def raw_env(self):
 
   def __getattr__(self, attr):
     """Delegate attribute access to underlying environment."""
-    return getattr(self._env, attr)
-
+    if "_env" in self.__dict__:
+      return getattr(self._env, attr)
+    return super().__getattribute__(attr)
 
 def _logarithmic_logging(episode: int,
                          ratios: Optional[Sequence[float]] = None) -> bool:
@@ -173,8 +174,9 @@ def step(self, action):
 
   def __getattr__(self, attr):
     """Delegate attribute access to underlying environment."""
-    return getattr(self._env, attr)
-
+    if "_env" in self.__dict__:
+      return getattr(self._env, attr)
+    return super().__getattribute__(attr)
 
 def _small_state_to_image(shape: Sequence[int],
                           observation: np.ndarray) -> np.ndarray:
@@ -307,8 +309,9 @@ def bsuite_info(self) -> Dict[str, Any]:
 
   def __getattr__(self, attr):
     """Delegate attribute access to underlying environment."""
-    return getattr(self._env, attr)
-
+    if "_env" in self.__dict__:
+      return getattr(self._env, attr)
+    return super().__getattribute__(attr)
 
 class RewardScale(environments.Environment):
   """Reward Scale environment wrapper."""
@@ -370,4 +373,6 @@ def bsuite_info(self) -> Dict[str, Any]:
 
   def __getattr__(self, attr):
     """Delegate attribute access to underlying environment."""
-    return getattr(self._env, attr)
+    if "_env" in self.__dict__:
+      return getattr(self._env, attr)
+    return super().__getattribute__(attr)