Role-Playing Ability on Role-playing
4.99CUS ScorePPR-GDE
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| PPR-GDEBackbone=Qwen2.5-3B-Instruct2026.05 | 4.99 | 2.6 | 4.93 | 4.17 | |
| PPR-GDEBackbone=Qwen3-0.6B2026.05 | 4.97 | 2.25 | 4.94 | 4.05 | |
| PPR-GDEBackbone=Qwen2.5-1.5B-Instruct2026.05 | 4.94 | 2.82 | 4.86 | 4.21 | |
| GRPOBackbone=Qwen2.5-3B-Instruct2026.05 | 4.88 | 2.51 | 4.79 | 4.06 | |
| PPOBackbone=Qwen3-0.6B2026.05 | 4.83 | 2.62 | 4.79 | 4.08 | |
| GRPOBackbone=Qwen2.5-1.5B-Instruct2026.05 | 4.81 | 2.44 | 4.66 | 3.97 | |
| GRPOBackbone=Qwen3-0.6B2026.05 | 4.8 | 2.99 | 4.66 | 4.15 | |
| PPOBackbone=Qwen2.5-1.5B-Instruct2026.05 | 4.77 | 2.16 | 4.63 | 3.85 | |
| PPOBackbone=Qwen2.5-3B-Instruct2026.05 | 4.71 | 2.74 | 4.72 | 4.06 | |
| w/o RLBackbone=Qwen2.5-3B-Instruct2026.05 | 4.04 | 3.11 | 4.2 | 3.78 | |
| w/o RLBackbone=Qwen3-0.6B2026.05 | 3.38 | 1.79 | 3.35 | 2.84 | |
| w/o RLBackbone=Qwen2.5-1.5B-Instruct2026.05 | 3.21 | 3 | 3.49 | 3.23 |