xpag.agents.rljax_agents.algorithm.base_class.actor_critic.OffPolicyActorCritic#

class OffPolicyActorCritic(num_agent_steps, observation_dim, action_dim, seed, max_grad_norm, gamma, nstep, num_critics, buffer_size, use_per, batch_size, start_steps, update_interval, update_interval_target=None, tau=None)#

Bases: ActorCriticMixIn, OffPolicyAlgorithm

Base class for off-policy Actor-Critic algorithms.

Methods

`calculate_value`	rtype: `Array`
`explore`
`get_key_list`
`get_mask`
`is_update`
`load_params`
`save_params`
`select_action`
`step`
`update`

Attributes

`kwargs_actor`
`kwargs_critic`
`name`