Role Generalization on RoleBench English 1.0
60.2CUS ScoreRoleGPT
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| RoleGPT2023.10 | 60.2 | 53.2 | 29.9 | 47.8 | |
| RoleLLaMA-7BModel size=7B2023.10 | 41.3 | 41.1 | 25.7 | 36 | |
| ChatPLUG2023.10 | 28.7 | 34.7 | 25 | 29.5 | |
| Alpaca2023.10 | 23.2 | 35.3 | 25.9 | 30.2 | |
| Vicuna2023.10 | 20.8 | 25.5 | 27.8 | 28.4 | |
| LLaMA-2-7B-Chat2023.10 | 20.8 | 26.2 | 20.2 | 22.4 | |
| LLaMA2023.10 | 13.2 | 12.3 | 25.5 | 22.4 |