Social Deduction Game Gameplay on Werewolf
29.1Village Team Win RateOurs + LSPO
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Ours + LSPOBackend LLM=Gemini-2.5-Flash2025.10 | 29.1 | 84.2 | 44.7 | |
| LSPOBackend LLM=Gemini-2.5-Flash2025.10 | 24.9 | 72.8 | 38.6 | |
| ReConBackend LLM=Gemini-2.5-Flash2025.10 | 24 | 66.7 | 36.3 | |
| Ours + ReActBackend LLM=Gemini-2.5-Flash2025.10 | 22.6 | 82.1 | 39.7 | |
| SLABackend LLM=Gemini-2.5-Flash2025.10 | 20.6 | 77.1 | 36.7 | |
| ReActBackend LLM=Gemini-2.5-Flash2025.10 | 17.5 | 79 | 35 |