def testServingEnvBadActions(self):
ev = CommonPolicyEvaluator(
env_creator=lambda _: SimpleServing(MockEnv(25)),
policy_graph=BadPolicyGraph,
sample_async=True,
batch_steps=40,
batch_mode="truncate_episodes")
self.assertRaises(Exception, lambda: ev.sample())
def testServingEnvOffPolicy(self):
ev = CommonPolicyEvaluator(
env_creator=lambda _: SimpleOffPolicyServing(MockEnv(25)),
policy_graph=MockPolicyGraph,
batch_steps=40,
batch_mode="complete_episodes")
for _ in range(3):
batch = ev.sample()
self.assertEqual(batch.count, 50)
def testServingEnvHorizonNotSupported(self):
ev = CommonPolicyEvaluator(
env_creator=lambda _: SimpleServing(MockEnv(25)),
policy_graph=MockPolicyGraph,
episode_horizon=20,
batch_steps=10,
batch_mode="complete_episodes")
ev.sample()
self.assertRaises(Exception, lambda: ev.sample())
def __init__(self, num, increment_obs=False):
if increment_obs:
# Observations are 0, 1, 2, 3... etc. as time advances
self.agents = [MockEnv2(5) for _ in range(num)]
else:
# Observations are all zeros
self.agents = [MockEnv(5) for _ in range(num)]
self.dones = set()
self.last_obs = {}
self.last_rew = {}
self.last_done = {}
self.last_info = {}
self.i = 0
self.num = num
self.observation_space = gym.spaces.Discrete(2)
self.action_space = gym.spaces.Discrete(2)
def __init__(self, num):
self.agents = [MockEnv(25) for _ in range(num)]
self.dones = set()
self.observation_space = gym.spaces.Discrete(2)
self.action_space = gym.spaces.Discrete(2)
