Reasoning accuracy on PRONTOQA 4-hop
85AccuracyAVI
Evaluation Results
| Method | Links | |
|---|---|---|
| AVIModel=Qwen 8B2025.05 | 85 | |
| AVIModel=Qwen 4B2025.05 | 72 | |
| No CorrectionModel=Qwen 8B2025.05 | 65 | |
| Self CorrectionModel=Qwen 4B2025.05 | 60 | |
| Self CorrectionModel=Qwen 8B2025.05 | 58 | |
| No CorrectionModel=Qwen 4B2025.05 | 52 |