add __all__

tianshou/core/losses.py

@@ -1,5 +1,11 @@
 import tensorflow as tf
 
+__all__ = [
+    'ppo_clip',
+    'REINFORCE',
+    'value_mse'
+]
+
 
 def ppo_clip(policy, clip_param):
     """

tianshou/core/opt.py

@@ -1,5 +1,9 @@
 import tensorflow as tf
 
+__all__ = [
+    'DPG',
+]
+
 
 def DPG(policy, action_value):
     """

tianshou/core/policy/base.py

@@ -1,7 +1,7 @@
 from __future__ import absolute_import
 from __future__ import division
 
-import tensorflow as tf
+__all__ = []
 
 
 class PolicyBase(object):

tianshou/core/policy/deterministic.py

@@ -1,10 +1,13 @@
 import tensorflow as tf
-import logging
 
 from .base import PolicyBase
 from ..random import OrnsteinUhlenbeckProcess
 from ..utils import identify_dependent_variables
 
+__all__ = [
+    'Deterministic',
+]
+
 
 class Deterministic(PolicyBase):
     """

tianshou/core/policy/distributional.py

@@ -1,8 +1,11 @@
 import tensorflow as tf
-import logging
 from .base import PolicyBase
 from ..utils import identify_dependent_variables
 
+__all__ = [
+    'Distributional',
+]
+
 
 class Distributional(PolicyBase):
     """

tianshou/core/policy/dqn.py

@@ -4,6 +4,10 @@ from .base import PolicyBase
 import tensorflow as tf
 import numpy as np
 
+__all__ = [
+    'DQN',
+]
+
 
 class DQN(PolicyBase):
     """

tianshou/core/random.py

@@ -5,6 +5,11 @@ adapted from keras-rl
 from __future__ import division
 import numpy as np
 
+__all__ = [
+    'GaussianWhiteNoiseProcess',
+    'OrnsteinUhlenbeckProcess',
+]
+
 
 class RandomProcess(object):
     """

tianshou/core/utils.py

@@ -1,5 +1,9 @@
 import tensorflow as tf
 
+__all__ = [
+    'get_soft_update_op',
+]
+
 
 def identify_dependent_variables(tensor, candidate_variables):
     """

tianshou/core/value_function/action_value.py

@@ -1,10 +1,14 @@
 from __future__ import absolute_import
-import logging
 import tensorflow as tf
 
 from .base import ValueFunctionBase
 from ..utils import identify_dependent_variables
 
+__all__ = [
+    'ActionValue',
+    'DQN',
+]
+
 
 class ActionValue(ValueFunctionBase):
     """

tianshou/core/value_function/base.py

@@ -2,6 +2,9 @@ from __future__ import absolute_import
 
 import tensorflow as tf
 
+__all__ = []
+
+
 class ValueFunctionBase(object):
     """
     Base class for value functions, including S-values and Q-values. The only

tianshou/core/value_function/state_value.py

@@ -1,11 +1,14 @@
 from __future__ import absolute_import
 
 import tensorflow as tf
-import logging
 
 from .base import ValueFunctionBase
 from ..utils import identify_dependent_variables
 
+__all__ = [
+    'StateValue',
+]
+
 
 class StateValue(ValueFunctionBase):
     """

tianshou/data/advantage_estimation.py

@@ -1,5 +1,12 @@
 import logging
-import numpy as np
+
+__all__ = [
+    'full_return',
+    'nstep_return',
+    'nstep_q_return',
+    'ddpg_return',
+]
+
 
 STATE = 0
 ACTION = 1

tianshou/data/data_buffer/base.py

@@ -1,3 +1,4 @@
+__all__ = []
 
 
 class DataBufferBase(object):

tianshou/data/data_buffer/batch_set.py

@@ -4,6 +4,11 @@ import logging
 
 from .base import DataBufferBase
 
+__all__ = [
+    'BatchSet'
+]
+
+
 STATE = 0
 ACTION = 1
 REWARD = 2

tianshou/data/data_buffer/replay_buffer_base.py

@@ -1,5 +1,8 @@
 from .base import DataBufferBase
 
+__all__ = []
+
+
 class ReplayBufferBase(DataBufferBase):
     """
     Base class for replay buffer.

tianshou/data/data_buffer/vanilla.py

@@ -3,6 +3,11 @@ import numpy as np
 
 from .replay_buffer_base import ReplayBufferBase
 
+__all__ = [
+    'VanillaReplayBuffer',
+]
+
+
 STATE = 0
 ACTION = 1
 REWARD = 2

tianshou/data/data_collector.py

@@ -7,6 +7,10 @@ from .data_buffer.batch_set import BatchSet
 from .utils import internal_key_match
 from ..core.policy.deterministic import Deterministic
 
+__all__ = [
+    'DataCollector',
+]
+
 
 class DataCollector(object):
     """

tianshou/data/tester.py

@@ -4,6 +4,10 @@ import gym
 import logging
 import numpy as np
 
+__all__ = [
+    'test_policy_in_env',
+]
+
 
 def test_policy_in_env(policy, env, num_timesteps=0, num_episodes=0,
                        discount_factor=0.99, seed=0, episode_cutoff=None):