Knowledge-Grounded Dialogue Generation (Fluency) on OpenDialKG
37.33Win RateRHO
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| RHOResponse Re-ranking (RR)=false, Comparison Baseline=GPT2+NPH2022.12 | 37.33 | 20.67 | 42 | |
| RHOResponse Re-ranking (RR)=true, Comparison Baseline=GPT2+NPH2022.12 | 32.33 | 16 | 51.67 | |
| RHOResponse Re-ranking (RR)=false, Comparison Baseline=BART+NPH2022.12 | 24.67 | 18.67 | 56.67 | |
| RHOResponse Re-ranking (RR)=true, Comparison Baseline=BART+NPH2022.12 | 17 | 12.67 | 70.33 |