def test_discrete_awr(observation_shape, action_size, scaler):
awr = DiscreteAWR(
batch_size=100,
batch_size_per_update=30,
n_actor_updates=1,
n_critic_updates=1,
scaler=scaler,
)
algo_tester(awr, observation_shape, state_value=True)
algo_update_tester(awr, observation_shape, action_size, True)
def test_discrete_awr_performance():
awr = DiscreteAWR(n_epochs=1)
algo_cartpole_tester(awr, n_trials=3)
def test_discrete_awr(observation_shape, action_size, scaler):
awr = DiscreteAWR(scaler=scaler, batch_size=100, batch_size_per_update=20)
algo_tester(awr, observation_shape, state_value=True)
algo_update_tester(awr, observation_shape, action_size, True)