MultiWOZ
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
Multiwoz
3.453G-Eval Score
6
MultiWoz NLG (test)
0.929BERTScore
6
MultiWoZ 2.1 (dev)
58.46Joint Goal Accuracy
6
MultiWOZ 89 scenarios
92.7Collaboration SR
5
MultiWOZ 16th cross-dataset task 2.4
42.84JGA (Attraction)
5
MultiWOZ 2.4 (5%)
54.09Joint Goal Acc
5
MultiWOZ 20% 2.0 (train)
47.89Joint Goal Accuracy
5
MultiWOZ 10% 2.0 (train)
45.1Joint Goal Acc
5
MultiWOZ 5% 2.0 (train)
40.9Joint Goal Acc
5
MultiWOZ 1% 2.0 (train)
29.72Joint Goal Accuracy
5
MultiWOZ (Human Evaluation)
91.67Success Rate
5
MultiWOZ High-Entropy SOP 2.2
94.2Vanilla APA
4
MultiWOZ Shallow Retrieval 2.2
98Vanilla APA
4
MultiWOZ Real World User Experiment
84.7Success Rate
4
MultiWOZ 2.4 (1%)
43.82Joint Goal Accuracy
4
MultiWOZ 2.1
86.3Inform Rate
4
MultiWOZ Restaurant domain 2.1
84Inform Rate
4
MultiWOZ Hotel domain 2.1
75Inform Rate
4
MultiWOZ 2.1 (train)
0.813Inform Rate
4
MultiWOZ Attraction domain 2.1
86Inform Rate
4
MultiWOZ (test)
77.5Precision
4
MultiWoZ Synthetic 2.1 (test)
76.2Purity
3
MultiWOZ 2.0 (test)
1.92Understanding
3
MultiWOZ ConvLab simulator
89.4Completion Rate
3
MultiWOZ human evaluation (test)
3.65Diversity
3