Role-Playing Evaluation (Conversational-Naturalness) on CN
65Win RateEBM
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| EBMOpponent=haruhi2026.05 | 65 | 20.8 | 14.2 | 44.2 | 75.8 | 1.47 | |
| EBMOpponent=qwen2.5-vl-7b2026.05 | 55.8 | 29.8 | 14.4 | 26 | 65.2 | 3.32 | |
| EBMOpponent=RoleMRC2026.05 | 54.4 | 28.4 | 17.2 | 26 | 65.7 | 1.64 | |
| EBMOpponent=intervl8b2026.05 | 53.4 | 29.2 | 17.4 | 24.2 | 64.6 | 2.74 | |
| EBMOpponent=qwen2.5-vl-7b-sft2026.05 | 50.2 | 31.8 | 18 | 18.4 | 61.2 | 6.42 | |
| EBMOpponent=Internvl14b2026.05 | 48 | 31.6 | 20.4 | 16.4 | 60.3 | 4.63 | |
| EBMOpponent=Qwen32b2026.05 | 47 | 44 | 9 | 3 | 51.6 | 0.5117 | |
| EBMOpponent=crab2026.05 | 41.2 | 40 | 18.8 | 1.2 | 50.7 | 0.8041 | |
| EBMOpponent=Internvl38b2026.05 | 40.2 | 42.4 | 17.4 | -2.2 | 48.7 | 0.6227 |