Role-playing Evaluation on RoleChat (val)
0.21Overall MSERoleJudge
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| RoleJudge2026.04 | 0.21 | 0.81 | 0.62 | |
| Gemini3 Pro2026.04 | 0.68 | 0.68 | 0.59 | |
| GPT-4o-audio2026.04 | 1.42 | 0.51 | 0.46 | |
| Qwen3-Omni2026.04 | 1.86 | 0.44 | 0.38 | |
| Qwen2-Audio2026.04 | 2.94 | 0.35 | 0.26 |