Role-playing Multiple-choice Evaluation on RoleAgentBench
51.02Friends AccuracySparse
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| SparseBackbone=Llama-3-8B2026.02 | 51.02 | 53.97 | 60.53 | 61.76 | 70.83 | |
| WandaBackbone=Llama-3-8B2026.02 | 45.91 | 54.76 | 63.16 | 55.88 | 62.5 | |
| SparseBackbone=Llama-2-13B2026.02 | 41.84 | 50 | 55.26 | 50 | 54.17 | |
| WandaBackbone=Llama-2-13B2026.02 | 39.8 | 48.41 | 52.63 | 52.94 | 50 | |
| RagBackbone=Llama-3-8B2026.02 | 30.61 | 47.62 | 50 | 34.09 | 45.83 | |
| RagBackbone=Llama-2-13B2026.02 | 23.47 | 45.24 | 44.74 | 34.09 | 45.83 | |
| PromptBackbone=Llama-2-13B2026.02 | 18.37 | 42.06 | 42.11 | 29.55 | 41.67 | |
| PromptBackbone=Llama-3-8B2026.02 | 18.37 | 42.06 | 42.11 | 29.55 | 41.67 |