def test_does_not_support_dict_obs_space(self):
"""Test that policy raises error if passed a dict obs space."""
env = GymEnv(DummyDictEnv(act_space_type='discrete'))
with pytest.raises(ValueError):
qf = SimpleQFunction(env.spec,
name='does_not_support_dict_obs_space')
DiscreteQFArgmaxPolicy(env_spec=env.spec, qf=qf)
def setup_method(self):
super().setup_method()
self.env = GymEnv(DummyDiscreteEnv())
self.qf = SimpleQFunction(self.env.spec)
self.policy = DiscreteQFArgmaxPolicy(env_spec=self.env.spec,
qf=self.qf)
self.sess.run(tf.compat.v1.global_variables_initializer())
self.env.reset()
def setUp(self):
super().setUp()
self.env = TfEnv(DummyDiscreteEnv())
self.qf = SimpleQFunction(self.env.spec)
self.policy = DiscreteQfDerivedPolicy(
env_spec=self.env.spec, qf=self.qf)
self.sess.run(tf.global_variables_initializer())
self.env.reset()
def test_invalid_action_spaces(self):
"""Test that policy raises error if passed a dict obs space."""
env = GymEnv(DummyDictEnv(act_space_type='box'))
with pytest.raises(ValueError):
qf = SimpleQFunction(env.spec)
DiscreteQFArgmaxPolicy(env_spec=env.spec, qf=qf)