xpag.agents.rljax_agents.algorithm.sac.SAC#

class SAC(num_agent_steps, observation_dim, action_dim, seed, max_grad_norm=None, gamma=0.99, nstep=1, num_critics=2, buffer_size=1000000, use_per=False, batch_size=256, start_steps=10000, update_interval=1, tau=0.005, fn_actor=None, fn_critic=None, lr_actor=0.0003, lr_critic=0.0003, lr_alpha=0.0003, units_actor=(256, 256), units_critic=(256, 256), log_std_min=-20.0, log_std_max=2.0, d2rl=False, init_alpha=1.0, adam_b1_alpha=0.9, *args, **kwargs)#

Bases: OffPolicyActorCritic

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