Update gym version.

PiperOrigin-RevId: 253615467
Change-Id: I142eaedd5c73627900fc2bc5918716ba15274a4f

setup.py

@@ -136,9 +136,9 @@ def main(_):
 
 TEST_REQUIRED_PACKAGES = [
     'atari_py == 0.1.7',
-    'gym == 0.10.11',
+    'gym == 0.12.5',
     'opencv-python >= 3.4.1.15',
-    'pybullet == 2.4.2',
+    'pybullet',
     'scipy == 1.1.0',
 ]
 

tf_agents/environments/atari_wrappers.py

@@ -47,13 +47,13 @@ def __getattr__(self, name):
   def _generate_observation(self):
     return np.concatenate(self._frames, axis=2)
 
-  def _reset(self):
+  def reset(self):
     observation = self._env.reset()
     for _ in range(FrameStack4.STACK_SIZE):
       self._frames.append(observation)
     return self._generate_observation()
 
-  def _step(self, action):
+  def step(self, action):
     observation, reward, done, info = self._env.step(action)
     self._frames.append(observation)
     return self._generate_observation(), reward, done, info

tf_agents/environments/suite_atari.py

@@ -73,7 +73,7 @@ def load(environment_name,
   gym_spec = gym.spec(environment_name)
   gym_env = gym_spec.make()
 
-  if max_episode_steps is None and gym_spec.timestep_limit is not None:
+  if max_episode_steps is None and gym_spec.max_episode_steps is not None:
     max_episode_steps = gym_spec.max_episode_steps
 
   return suite_gym.wrap_env(

tf_agents/environments/suite_gym.py

@@ -46,7 +46,7 @@ def load(environment_name,
     discount: Discount to use for the environment.
     max_episode_steps: If None the max_episode_steps will be set to the default
       step limit defined in the environment's spec. No limit is applied if set
-      to 0 or if there is no timestep_limit set in the environment's spec.
+      to 0 or if there is no max_episode_steps set in the environment's spec.
     gym_env_wrappers: Iterable with references to wrapper classes to use
       directly on the gym environment.
     env_wrappers: Iterable with references to wrapper classes to use on the
@@ -63,7 +63,7 @@ def load(environment_name,
   gym_spec = gym.spec(environment_name)
   gym_env = gym_spec.make()
 
-  if max_episode_steps is None and gym_spec.timestep_limit is not None:
+  if max_episode_steps is None and gym_spec.max_episode_steps is not None:
     max_episode_steps = gym_spec.max_episode_steps
 
   return wrap_env(
@@ -93,7 +93,7 @@ def wrap_env(gym_env,
     gym_env: An instance of OpenAI gym environment.
     discount: Discount to use for the environment.
     max_episode_steps: Used to create a TimeLimitWrapper. No limit is applied
-      if set to 0. Usually set to `gym_spec.timestep_limit` as done in `load.
+      if set to 0. Usually set to `gym_spec.max_episode_steps` as done in `load.
     gym_env_wrappers: Iterable with references to wrapper classes to use
       directly on the gym environment.
     time_limit_wrapper: Wrapper that accepts (env, max_episode_steps) params to