Reasoning accuracy on PRONTOQA 5-hop
97.8AccuracyLogicAgent
Evaluation Results
| Method | Links | |
|---|---|---|
| LogicAgentType=AR+SR, Base Model=Qwen2.52025.09 | 97.8 | |
| SymbCoTType=SR, Base Model=Qwen2.52025.09 | 95.2 | |
| AristotleType=AR+SR, Base Model=Qwen2.52025.09 | 94.8 | |
| CoTType=LR, Base Model=Qwen2.52025.09 | 92.4 | |
| Logic-LMType=SR, Base Model=Qwen2.52025.09 | 91.89 | |
| ToTType=AR, Base Model=Qwen2.52025.09 | 82.5 | |
| NaiveType=LR, Base Model=Qwen2.52025.09 | 82 | |
| AVIModel=Qwen 8B2025.05 | 81 | |
| CRType=AR, Base Model=Qwen2.52025.09 | 80.2 | |
| AVIModel=Qwen 4B2025.05 | 77 | |
| No CorrectionModel=Qwen 4B2025.05 | 59 | |
| No CorrectionModel=Qwen 8B2025.05 | 52 | |
| Self CorrectionModel=Qwen 4B2025.05 | 48 | |
| Self CorrectionModel=Qwen 8B2025.05 | 46 |