Open-ended Dialogue Generation on CMD (test)
19.4Rouge-LNSR
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| NSR2026.05 | 19.4 | 79.2 | 22 | 87.9 | 52.1 | |
| W-REINFORCEpositive sample weight (w+)=0.05, negative sample weight (w-)=12026.05 | 19.4 | 77.8 | 22 | 87.4 | 51.7 | |
| NSR2026.05 | 19.4 | 79.2 | 22 | 87.9 | 52.1 | |
| W-REINFORCEw+=0.05, w-=12026.05 | 19.4 | 77.8 | 22 | 87.4 | 51.7 | |
| W-REINFORCEpositive sample weight (w+)=1, negative sample weight (w-)=52026.05 | 18.7 | 79.3 | 21 | 86.2 | 51.3 | |
| EAPO2026.05 | 18.7 | 79.8 | 21.3 | 86.4 | 51.6 | |
| W-REINFORCEw+=1, w-=52026.05 | 18.7 | 79.3 | 21 | 86.2 | 51.3 | |
| EAPO2026.05 | 18.7 | 79.8 | 21.3 | 86.4 | 51.6 | |
| GRPO2026.05 | 18.6 | 78.3 | 20.6 | 83.8 | 50.3 | |
| W-REINFORCEpositive sample weight (w+)=0.2, negative sample weight (w-)=12026.05 | 18.6 | 75.4 | 21.6 | 87.6 | 50.8 | |
| GRPO2026.05 | 18.6 | 78.3 | 20.6 | 83.8 | 50.3 | |
| W-REINFORCEw+=0.2, w-=12026.05 | 18.6 | 75.4 | 21.6 | 87.6 | 50.8 | |
| W-REINFORCEpositive sample weight (w+)=0.1, negative sample weight (w-)=12026.05 | 18.5 | 80.8 | 20.7 | 87.1 | 51.8 | |
| W-REINFORCEw+=0.1, w-=12026.05 | 18.5 | 80.8 | 20.7 | 87.1 | 51.8 | |
| W-REINFORCEpositive sample weight (w+)=1, negative sample weight (w-)=12026.05 | 18.4 | 78.7 | 20.7 | 85.2 | 50.8 | |
| W-REINFORCEw+=1, w-=12026.05 | 18.4 | 78.7 | 20.7 | 85.2 | 50.8 | |
| W-REINFORCEpositive sample weight (w+)=5, negative sample weight (w-)=12026.05 | 17.3 | 75.2 | 19.9 | 83.8 | 49 | |
| W-REINFORCEw+=5, w-=12026.05 | 17.3 | 75.2 | 19.9 | 83.8 | 49 | |
| W-REINFORCEpositive sample weight (w+)=1, negative sample weight (w-)=0.22026.05 | 17 | 76.3 | 19.5 | 85.2 | 49.5 | |
| W-REINFORCEw+=1, w-=0.22026.05 | 17 | 76.3 | 19.5 | 85.2 | 49.5 | |
| W-REINFORCEpositive sample weight (w+)=1, negative sample weight (w-)=0.052026.05 | 16 | 74.9 | 18.4 | 82.7 | 48 | |
| W-REINFORCEw+=1, w-=0.052026.05 | 16 | 74.9 | 18.4 | 82.7 | 48 | |
| W-REINFORCEpositive sample weight (w+)=1, negative sample weight (w-)=0.12026.05 | 15.8 | 72.4 | 18.3 | 80.6 | 46.7 | |
| W-REINFORCEw+=1, w-=0.12026.05 | 15.8 | 72.4 | 18.3 | 80.6 | 46.7 | |
| PSR2026.05 | 14.7 | 68.8 | 17.3 | 82.1 | 45.7 | |
| PSR2026.05 | 14.7 | 68.8 | 17.3 | 82.1 | 45.7 |