Dialogue Response Generation on 100 randomly sampled conversational pairs (test)
66.1AppropriatenessSaBART
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| SaBARTComparison Partner=ConceptFlow2023.06 | 66.1 | 70.2 | — | |
| SaBARTComparison Partner=SaBART - w/o st-agg2023.06 | 62.3 | 60 | — | |
| SaBARTComparison Partner=SaBART - w/o dy-agg2023.06 | 61.9 | 65 | — | |
| SaBART - w/o dy-aggComparison Partner=SaBART2023.06 | 38.1 | 35 | — | |
| SaBART - w/o st-aggComparison Partner=SaBART2023.06 | 37.7 | 40 | — | |
| ConceptFlowComparison Partner=SaBART2023.06 | 33.9 | 29.8 | — |