Microsoft Dialogue Challenge: End-To-End Task-Completion Dialogue Challenge

Movie Leaderboard

Rank Model Success Rate (Simulation) Success Rate (Human) Rating (Human)

1

Oct 25, 2018
Double Q

National Taiwan Unisversity

41.8% 31.1% 2.65/5

1

Sep 20, 2018
DQN

single model

44.1% 30.8% 2.62/5

3

Oct 25, 2018
HDQN

National Taiwan Unisversity

33.3% 27.3% 2.49/5

4

Oct 25, 2018
Transfer-DDQ

Beijing University of Posts and Telecommunications

11.5% 9.66% 2.24/5

5

Sep 20, 2018
Rule

single model

6.13% 6.42% 1.78/5

Restaurant Leaderboard

Rank Model Success Rate (Simulation) Success Rate (Human) Rating (Human)

1

Sep 20, 2018
DQN

single model

30.18% 22.9% 2.35/5

2

Sep 20, 2018
Rule

single model

7.22% 6.85% 1.94/5

Taxi Leaderboard

Rank Model Success Rate (Simulation) Success Rate (Human) Rating (Human)

1

Sep 20, 2018
DQN

single model

43.5% 25.2% 2.38/5

2

Sep 20, 2018
Rule

single model

12.2% 8.70% 1.71/5