Table 7. Results obtained by different dialogue agents.

Agent	Success rate	Avg. Reward	Avg. Dialogue Length
Random Agent	0.011	-232	15.25
Rule Agent	0.000	-127	11.00
ACTR Agent	0.6615	36.11	11.32
ACTR+SR	0.6753	37.18	11.20
Vanilla DQNTR Agent	0.8413	68.41	9.30
Vanilla DQNTR+SR Agent	0.8573	70.82	8.91
DDQNTR Agent	0.8693	73.02	8.68
${\text{DDQN}}_{TR+SR}^B$ Agent	0.8610	72.14	8.52
${\text{DDQN}}_{TR+SR}^{XL}$ Agent	0.8904	79.86	8.38