def test_mopo_impl(
observation_shape,
action_size,
learning_rate,
optim_factory,
encoder_factory,
n_ensembles,
lam,
discrete_action,
scaler,
):
impl = MOPOImpl(
observation_shape,
action_size,
learning_rate,
optim_factory,
encoder_factory,
n_ensembles,
lam,
discrete_action,
use_gpu=False,
scaler=scaler,
)
impl.build()
torch_impl_tester(impl, discrete=discrete_action)
def test_mopo_impl(
observation_shape,
action_size,
learning_rate,
optim_factory,
encoder_factory,
n_ensembles,
lam,
discrete_action,
scaler,
action_scaler,
):
impl = MOPOImpl(
observation_shape=observation_shape,
action_size=action_size,
learning_rate=learning_rate,
optim_factory=optim_factory,
encoder_factory=encoder_factory,
n_ensembles=n_ensembles,
lam=lam,
discrete_action=discrete_action,
use_gpu=None,
scaler=scaler,
action_scaler=action_scaler if not discrete_action else None,
)
impl.build()
torch_impl_tester(impl, discrete=discrete_action)
def test_probabilistic_ensemble_dynamics_impl(
observation_shape,
action_size,
learning_rate,
optim_factory,
encoder_factory,
n_ensembles,
variance_type,
discrete_action,
scaler,
action_scaler,
reward_scaler,
):
impl = ProbabilisticEnsembleDynamicsImpl(
observation_shape=observation_shape,
action_size=action_size,
learning_rate=learning_rate,
optim_factory=optim_factory,
encoder_factory=encoder_factory,
n_ensembles=n_ensembles,
variance_type=variance_type,
discrete_action=discrete_action,
use_gpu=None,
scaler=scaler,
action_scaler=action_scaler if not discrete_action else None,
reward_scaler=reward_scaler,
)
impl.build()
torch_impl_tester(impl, discrete=discrete_action, n_ensembles=n_ensembles)
def test_mopo_impl(observation_shape, action_size, learning_rate, eps,
weight_decay, n_ensembles, lam, use_batch_norm,
discrete_action, scaler):
impl = MOPOImpl(observation_shape,
action_size,
learning_rate,
eps,
weight_decay,
n_ensembles,
lam,
use_batch_norm,
discrete_action,
use_gpu=False,
scaler=scaler)
torch_impl_tester(impl, discrete=discrete_action)