Response Generation on DailyDialog (test)
35.4BLEU-2Hier
Evaluation Results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Hierbeam size=52021.05 | 35.4 | 41.3 | 33.2 | 32.1 | 6.5 | 32.3 | — | — | — | — | — | |
| Dualbeam size=52021.05 | 35 | 40.8 | 32.7 | 31.5 | 6.6 | 33 | — | — | — | — | — | |
| PLATO w/o Lbeam size=5, pre-trained=true, external corpus for pretraining=true2021.05 | 32.2 | 40.5 | — | — | 4.6 | 24.6 | — | — | — | — | — | |
| Transformerbeam size=52021.05 | 31.7 | 38.3 | 29.1 | 27.8 | 5.8 | 30.5 | — | — | — | — | — | |
| PLATObeam size=5, pre-trained=true, external corpus for pretraining=true, source of results=Bao et al. (2020)2021.05 | 31.1 | 39.7 | — | — | 5.3 | 29.1 | — | — | — | — | — | |
| BART + DialoGPSTraining=fine-tuning, Beam search width=52023.06 | 29.18 | — | — | 12.05 | — | — | — | — | 15.3 | 9.73 | — | |
| DialoFlowModel Size=large, Decoding Strategy=beam2021.06 | 28.02 | — | — | 11.57 | — | — | 3.86 | 4.08 | 17 | 9.87 | 12.08 | |
| DialoFlowTraining=fine-tuning, Beam search width=52023.06 | 28.02 | — | — | 11.57 | — | — | — | — | 16.4 | 9.46 | — | |
| BARTTraining=fine-tuning, Beam search width=52023.06 | 27.87 | — | — | 10.85 | — | — | — | — | 14.69 | 9.29 | — | |
| DialoFlowModel Size=medium, Decoding Strategy=beam2021.06 | 27.63 | — | — | 11.33 | — | — | 3.8 | 4.02 | 16.7 | 9.83 | 12.06 | |
| Seq2Seqbeam size=5, source of results=Bao et al. (2020)2021.05 | 26.8 | 33.6 | — | — | 3 | 12.8 | — | — | — | — | — | |
| DialoFlowModel Size=base, Decoding Strategy=beam2021.06 | 26.47 | — | — | 10.12 | — | — | 3.65 | 3.84 | 16.1 | 9.62 | 12 | |
| DialoGPTModel Size=medium, Decoding Strategy=beam2021.06 | 25.39 | — | — | 9.99 | — | — | 3.47 | 3.65 | 15.9 | 9.64 | 12.88 | |
| iVAE_MIbeam size=52021.05 | 24.9 | 30.9 | — | — | 2.9 | 25 | — | — | — | — | — | |
| DialoGPTModel Size=large, Decoding Strategy=beam2021.06 | 23.69 | — | — | 9.2 | — | — | 3.3 | 3.46 | 15.7 | 9.78 | 13.24 | |
| DialoGPTModel Size=base, Decoding Strategy=beam2021.06 | 18.83 | — | — | 6.63 | — | — | 2.28 | 2.78 | 15.5 | 9.8 | 18.82 |