def initialize_model(self, states_spec, actions_spec, config):
return QNstepModel(
states_spec=states_spec,
actions_spec=actions_spec,
network_spec=self.network_spec,
config=config
)
def initialize_model(self):
return QNstepModel(
states_spec=self.states_spec,
actions_spec=self.actions_spec,
network_spec=self.network_spec,
device=self.device,
session_config=self.session_config,
scope=self.scope,
saver_spec=self.saver_spec,
summary_spec=self.summary_spec,
distributed_spec=self.distributed_spec,
optimizer=self.optimizer,
discount=self.discount,
variable_noise=self.variable_noise,
states_preprocessing_spec=self.states_preprocessing_spec,
explorations_spec=self.explorations_spec,
reward_preprocessing_spec=self.reward_preprocessing_spec,
distributions_spec=self.distributions_spec,
entropy_regularization=self.entropy_regularization,
target_sync_frequency=self.target_sync_frequency,
target_update_weight=self.target_update_weight,
double_q_model=self.double_q_model,
huber_loss=self.huber_loss,
# TEMP: Random sampling fix
random_sampling_fix=False)
def initialize_model(self, states_spec, actions_spec):
return QNstepModel(states_spec=states_spec,
actions_spec=actions_spec,
network_spec=self.network_spec,
device=self.device,
scope=self.scope,
saver_spec=self.saver_spec,
summary_spec=self.summary_spec,
distributed_spec=self.distributed_spec,
optimizer=self.optimizer,
discount=self.discount,
normalize_rewards=self.normalize_rewards,
variable_noise=self.variable_noise,
distributions_spec=self.distributions_spec,
entropy_regularization=self.entropy_regularization,
target_sync_frequency=self.target_sync_frequency,
target_update_weight=self.target_update_weight,
double_q_model=self.double_q_model,
huber_loss=self.huber_loss)
def initialize_model(self):
return QNstepModel(states=self.states,
actions=self.actions,
scope=self.scope,
device=self.device,
saver=self.saver,
summarizer=self.summarizer,
distributed=self.distributed,
batching_capacity=self.batching_capacity,
variable_noise=self.variable_noise,
states_preprocessing=self.states_preprocessing,
actions_exploration=self.actions_exploration,
reward_preprocessing=self.reward_preprocessing,
update_mode=self.update_mode,
memory=self.memory,
optimizer=self.optimizer,
discount=self.discount,
network=self.network,
distributions=self.distributions,
entropy_regularization=self.entropy_regularization,
target_sync_frequency=self.target_sync_frequency,
target_update_weight=self.target_update_weight,
double_q_model=self.double_q_model,
huber_loss=self.huber_loss)