added metadata and reward_range for each tasks

envs/atari.py

@@ -4,7 +4,7 @@ import numpy as np
 
 class Atari:
     LOCK = None
-
+    metadata = {}
     def __init__(
         self,
         name,
@@ -64,6 +64,7 @@ class Atari:
         self._last_lives = None
         self._done = True
         self._step = 0
+        self.reward_range = [-np.inf, np.inf]
 
     @property
     def observation_space(self):

envs/crafter.py

@@ -3,12 +3,14 @@ import numpy as np
 
 
 class Crafter:
+    metadata = {}
     def __init__(self, task, size=(64, 64), seed=None):
         assert task in ("reward", "noreward")
         import crafter
 
         self._env = crafter.Env(size=size, reward=(task == "reward"), seed=seed)
         self._achievements = crafter.constants.achievements.copy()
+        self.reward_range = [-np.inf, np.inf]
 
     @property
     def observation_space(self):

envs/dmc.py

@@ -3,6 +3,7 @@ import numpy as np
 
 
 class DeepMindControl:
+    metadata = {}
     def __init__(self, name, action_repeat=1, size=(64, 64), camera=None):
         domain, task = name.split("_", 1)
         if domain == "cup":  # Only domain with multiple words.
@@ -19,6 +20,7 @@ class DeepMindControl:
         if camera is None:
             camera = dict(quadruped=2).get(domain, 0)
         self._camera = camera
+        self.reward_range = [-np.inf, np.inf]
 
     @property
     def observation_space(self):