def make_dqn_agent(self, env, q_func, opt, explorer, rbuf, gpu):
return double_iqn.DoubleIQN(
q_func,
opt,
rbuf,
gpu=gpu,
gamma=0.9,
explorer=explorer,
replay_start_size=100,
target_update_interval=100,
quantile_thresholds_N=self.quantile_thresholds_N,
quantile_thresholds_N_prime=self.quantile_thresholds_N_prime,
)
def make_dqn_agent(self, env, q_func, opt, explorer, rbuf, gpu):
return double_iqn.DoubleIQN(
q_func,
opt,
rbuf,
gpu=gpu,
gamma=0.9,
explorer=explorer,
replay_start_size=100,
target_update_interval=100,
quantile_thresholds_N=32,
quantile_thresholds_N_prime=32,
recurrent=True,
act_deterministically=True,
)