from .base import AlgoBase
from .torch.sac_impl import SACImpl
[docs]class SAC(AlgoBase):
""" Soft Actor-Critic algorithm.
SAC is a DDPG-based maximum entropy RL algorithm, which produces
state-of-the-art performance in online RL settings.
SAC leverages twin Q functions proposed in TD3. Additionally,
`delayed policy update` in TD3 is also implemented, which is not done in
the paper.
.. math::
L(\\theta_i) = \\mathbb{E}_{s_t, a_t, r_{t+1}, s_{t+1} \\sim D,
a_{t+1} \\sim \\pi_\\phi(\\cdot|s_{t+1})} [
(y - Q_{\\theta_i}(s_t, a_t))^2]
.. math::
y = r_{t+1} + \\gamma (\\min_j Q_{\\theta_j}(s_{t+1}, a_{t+1})
- \\alpha \\log (\\pi_\\phi(a_{t+1}|s_{t+1})))
.. math::
J(\\phi) = \\mathbb{E}_{s_t \\sim D, a_t \\sim \\pi_\\phi(\\cdot|s_t)}
[\\alpha \\log (\\pi_\\phi (a_t|s_t))
- \\min_i Q_{\\theta_i}(s_t, \\pi_\\phi(a_t|s_t))]
The temperature parameter :math:`\\alpha` is also automatically adjustable.
.. math::
J(\\alpha) = \\mathbb{E}_{s_t \\sim D, a_t \\sim \\pi_\\phi\(\cdot|s_t)}
[-\\alpha (\\log (\\pi_\\phi(a_t|s_t)) + H)]
where :math:`H` is a target
entropy, which is defined as :math:`\dim a`.
References:
* `Haarnoja et al., Soft Actor-Critic: Off-Policy Maximum Entropy Deep
Reinforcement Learning with a Stochastic Actor.
<https://arxiv.org/abs/1801.01290>`_
* `Haarnoja et al., Soft Actor-Critic Algorithms and Applications.
<https://arxiv.org/abs/1812.05905>`_
Args:
actor_learning_rate (float): learning rate for policy function.
critic_learning_rate (float): learning rate for Q functions.
temp_learning_rate (float): learning rate for temperature parameter.
batch_size (int): mini-batch size.
n_frames (int): the number of frames to stack for image observation.
gamma (float): discount factor.
tau (float): target network synchronization coefficiency.
n_critics (int): the number of Q functions for ensemble.
bootstrap (bool): flag to bootstrap Q functions.
share_encoder (bool): flag to share encoder network.
update_actor_interval (int): interval to update policy function.
initial_temperature (float): initial temperature value.
eps (float): :math:`\\epsilon` for Adam optimizer.
use_batch_norm (bool): flag to insert batch normalization layers.
q_func_type (str): type of Q function. Available options are
`['mean', 'qr', 'iqn', 'fqf']`.
use_gpu (bool, int or d3rlpy.gpu.Device):
flag to use GPU, device ID or device.
scaler (d3rlpy.preprocessing.Scaler or str): preprocessor.
The available options are `['pixel', 'min_max', 'standard']`
augmentation (d3rlpy.augmentation.AugmentationPipeline or list(str)):
augmentation pipeline.
n_augmentations (int): the number of data augmentations to update.
encoder_params (dict): optional arguments for encoder setup. If the
observation is pixel, you can pass ``filters`` with list of tuples
consisting with ``(filter_size, kernel_size, stride)`` and
``feature_size`` with an integer scaler for the last linear layer
size. If the observation is vector, you can pass ``hidden_units``
with list of hidden unit sizes.
dynamics (d3rlpy.dynamics.base.DynamicsBase): dynamics model for data
augmentation.
impl (d3rlpy.algos.torch.sac_impl.SACImpl): algorithm implementation.
Attributes:
actor_learning_rate (float): learning rate for policy function.
critic_learning_rate (float): learning rate for Q functions.
temp_learning_rate (float): learning rate for temperature parameter.
batch_size (int): mini-batch size.
n_frames (int): the number of frames to stack for image observation.
gamma (float): discount factor.
tau (float): target network synchronization coefficiency.
n_critics (int): the number of Q functions for ensemble.
bootstrap (bool): flag to bootstrap Q functions.
share_encoder (bool): flag to share encoder network.
update_actor_interval (int): interval to update policy function.
initial_temperature (float): initial temperature value.
eps (float): :math:`\\epsilon` for Adam optimizer.
use_batch_norm (bool): flag to insert batch normalization layers.
q_func_type (str): type of Q function.
use_gpu (d3rlpy.gpu.Device): GPU device.
scaler (d3rlpy.preprocessing.Scaler): preprocessor.
augmentation (d3rlpy.augmentation.AugmentationPipeline):
augmentation pipeline.
n_augmentations (int): the number of data augmentations to update.
encoder_params (dict): optional arguments for encoder setup.
dynamics (d3rlpy.dynamics.base.DynamicsBase): dynamics model.
impl (d3rlpy.algos.torch.sac_impl.SACImpl): algorithm implementation.
eval_results_ (dict): evaluation results.
"""
def __init__(self,
*,
actor_learning_rate=3e-4,
critic_learning_rate=3e-4,
temp_learning_rate=3e-4,
batch_size=100,
n_frames=1,
gamma=0.99,
tau=0.005,
n_critics=2,
bootstrap=False,
share_encoder=False,
update_actor_interval=2,
initial_temperature=1.0,
eps=1e-8,
use_batch_norm=False,
q_func_type='mean',
use_gpu=False,
scaler=None,
augmentation=[],
n_augmentations=1,
encoder_params={},
dynamics=None,
impl=None,
**kwargs):
super().__init__(batch_size=batch_size,
n_frames=n_frames,
scaler=scaler,
augmentation=augmentation,
dynamics=dynamics,
use_gpu=use_gpu)
self.actor_learning_rate = actor_learning_rate
self.critic_learning_rate = critic_learning_rate
self.temp_learning_rate = temp_learning_rate
self.gamma = gamma
self.tau = tau
self.n_critics = n_critics
self.bootstrap = bootstrap
self.share_encoder = share_encoder
self.update_actor_interval = update_actor_interval
self.initial_temperature = initial_temperature
self.eps = eps
self.use_batch_norm = use_batch_norm
self.q_func_type = q_func_type
self.n_augmentations = n_augmentations
self.encoder_params = encoder_params
self.impl = impl
[docs] def create_impl(self, observation_shape, action_size):
self.impl = SACImpl(observation_shape=observation_shape,
action_size=action_size,
actor_learning_rate=self.actor_learning_rate,
critic_learning_rate=self.critic_learning_rate,
temp_learning_rate=self.temp_learning_rate,
gamma=self.gamma,
tau=self.tau,
n_critics=self.n_critics,
bootstrap=self.bootstrap,
share_encoder=self.share_encoder,
initial_temperature=self.initial_temperature,
eps=self.eps,
use_batch_norm=self.use_batch_norm,
q_func_type=self.q_func_type,
use_gpu=self.use_gpu,
scaler=self.scaler,
augmentation=self.augmentation,
n_augmentations=self.n_augmentations,
encoder_params=self.encoder_params)
self.impl.build()
[docs] def update(self, epoch, total_step, batch):
critic_loss = self.impl.update_critic(batch.observations,
batch.actions,
batch.next_rewards,
batch.next_observations,
batch.terminals)
# delayed policy update
if total_step % self.update_actor_interval == 0:
actor_loss = self.impl.update_actor(batch.observations)
temp_loss, temp = self.impl.update_temp(batch.observations)
self.impl.update_critic_target()
self.impl.update_actor_target()
else:
actor_loss = None
temp_loss = None
temp = None
return critic_loss, actor_loss, temp_loss, temp
def _get_loss_labels(self):
return ['critic_loss', 'actor_loss', 'temp_loss', 'temp']