def __init__(self, env, agent_params):
super(BCAgent, self).__init__()
# init vars
self.env = env
self.agent_params = agent_params
# actor/policy
self.actor = MLPPolicySL(
self.agent_params['ac_dim'],
self.agent_params['ob_dim'],
self.agent_params['n_layers'],
self.agent_params['size'],
discrete=self.agent_params['discrete'],
learning_rate=self.agent_params['learning_rate'],
siren=self.agent_params['siren'],
train_separate_offset=self.agent_params['train_separate_params'],
supervision_mode=self.agent_params['supervision_mode'],
offset_learning_rate=self.agent_params['offset_learning_rate'],
auto_cast=self.agent_params['auto_cast'],
gradient_loss_scale=self.agent_params['gradient_loss_scale'],
additional_activation=self.agent_params['additional_activation'],
omega=self.agent_params['omega'])
# replay buffer
self.replay_buffer = ReplayBuffer(
self.agent_params['max_replay_buffer_size'],
epsilon_s=self.agent_params['epsilon_s'])
def __init__(self, env, agent_params):
super(BCAgent, self).__init__()
# init vars
self.env = env
self.agent_params = agent_params
# actor/policy
self.actor = MLPPolicySL(
self.agent_params['ac_dim'],
self.agent_params['ob_dim'],
self.agent_params['n_layers'],
self.agent_params['size'],
discrete=self.agent_params['discrete'],
learning_rate=self.agent_params['learning_rate'],
)
# replay buffer
self.replay_buffer = ReplayBuffer(self.agent_params['max_replay_buffer_size'])
def __init__(self, env, agent_params):
super(BCAgent, self).__init__()
# init vars
self.env = env
self.agent_params = agent_params
# actor/policy
self.actor = MLPPolicySL(
self.agent_params["ac_dim"],
self.agent_params["ob_dim"],
self.agent_params["n_layers"],
self.agent_params["size"],
discrete=self.agent_params["discrete"],
learning_rate=self.agent_params["learning_rate"],
)
# replay buffer
self.replay_buffer = ReplayBuffer(self.agent_params["max_replay_buffer_size"])