Dialogue Response Generation on SpokenWOZ
2.61GPT ScoreCT-Duplex
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| CT-Duplexchronological thinking=true2025.10 | 2.61 | 16.3 | 59 | |
| GT-LMdescription=optimal cascaded system2025.10 | 2.48 | 7.6 | 55 | |
| CT-Duplexchronological thinking=false2025.10 | 2.4 | 12.92 | 52 | |
| SALM-Duplex*reproduced=true2025.10 | 2.11 | 8.73 | 34 |