Improve docstrings

chainerrl/action_value.py

@@ -41,7 +41,7 @@ def evaluate_actions(self, actions):
 
 
 class DiscreteActionValue(ActionValue):
-    """Qfunction output for discrete action space.
+    """Q-function output for discrete action space.
 
     Args:
         q_values (ndarray or chainer.Variable):
@@ -95,7 +95,7 @@ def __repr__(self):
 
 
 class QuadraticActionValue(ActionValue):
-    """Qfunction output for continuous action space.
+    """Q-function output for continuous action space.
 
     See: http://arxiv.org/abs/1603.00748
 

chainerrl/v_functions/v_functions.py

@@ -16,11 +16,11 @@
 
 class SingleModelVFunction(
         chainer.Chain, VFunction, RecurrentChainMixin):
-    """Q-function with discrete actions.
+    """V-function
 
     Args:
         model (chainer.Link):
-            Link that is callable and outputs action values.
+            Link that is callable, inputs states, and outputs state values.
     """
 
     def __init__(self, model):

examples/gym/train_a3c_gym.py

@@ -1,6 +1,6 @@
 """An example of training A3C against OpenAI Gym Envs.
 
-This script is an example of training a PCL agent against OpenAI Gym envs.
+This script is an example of training a A3C agent against OpenAI Gym envs.
 Both discrete and continuous action spaces are supported.
 
 To solve CartPole-v0, run:

examples/gym/train_reinforce_gym.py

@@ -4,10 +4,10 @@
 envs. Both discrete and continuous action spaces are supported.
 
 To solve CartPole-v0, run:
-    python train_pcl_gym.py
+    python train_reinforce_gym.py
 
 To solve InvertedPendulum-v1, run:
-    python train_pcl_gym.py --env InvertedPendulum-v1
+    python train_reinforce_gym.py --env InvertedPendulum-v1
 """
 from __future__ import division
 from __future__ import print_function