Knowledge-grounded dialogue generation on OpenDialKG (test)
20.77BLEU-4RHO
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| RHOGlobal knowledge grounding (GKG)=true2022.12 | 20.77 | 39.54 | 40.65 | 48.41 | 43.84 | 97.2 | 45.63 | 67.4 | |
| RHOLocal knowledge grounding (LKG)=true, Global knowledge grounding (GKG)=true2022.12 | 20.63 | 39.51 | 45.96 | 50.35 | 46.03 | 98.26 | 50.74 | 71.47 | |
| RHOLocal knowledge grounding (LKG)=true2022.12 | 19.89 | 39.95 | 43.04 | 48.91 | 44.37 | 97.38 | 45.57 | 67.77 | |
| RHOLocal knowledge grounding (LKG)=true, Global knowledge grounding (GKG)=true, Response re-ranking (RR)=true2022.12 | 19.11 | 38.45 | 47.99 | 50.58 | 46.41 | 98.53 | 51.77 | 72.29 | |
| BART+NPH2022.12 | 15.53 | 34.99 | 42.41 | 47.94 | 43.56 | 96.44 | 44.12 | 65.98 | |
| BART2022.12 | 14.45 | 33.33 | 39 | 46.97 | 42.75 | 96.99 | 44.96 | 62.87 | |
| KG-BART2022.12 | 13.72 | 33.31 | 41.87 | 45.55 | 42.86 | 97.68 | 45.63 | 64.58 | |
| GPT2+NPH2022.12 | 10.41 | 29.93 | 40.83 | 47.45 | 42.45 | 95.61 | 33.39 | 53.96 | |
| GPT22022.12 | 10.27 | 29.59 | 39.6 | 46.86 | 42.07 | 91.62 | 33.26 | 52.3 | |
| EARL2022.12 | 7.97 | 23.61 | 39.93 | 37.88 | 35.59 | 86.61 | 45.17 | 64.44 |