Multi-Hop Question Answering on MuSiQue (Accuracy & Recall)
36.6AccuracyHyGRAG +Inference
Evaluation results
| Method | Links | ||
|---|---|---|---|
| HyGRAG +InferenceBackbone=Llama-3.1-8B-Instruct2026.06 | 36.6 | 48.36 | |
| HyGRAGBackbone=Llama-3.1-8B-Instruct2026.06 | 35.87 | 48.06 | |
| VanillaRAGBackbone=Llama-3.1-8B-Instruct2026.06 | 32.3 | 43.92 | |
| RAPTOR-KBackbone=Llama-3.1-8B-Instruct2026.06 | 31.67 | 43.91 | |
| RAPTORBackbone=Llama-3.1-8B-Instruct2026.06 | 31.17 | 43.47 | |
| L-LightRAGBackbone=Llama-3.1-8B-Instruct2026.06 | 30.63 | 43.66 | |
| HippoRAG2Backbone=Llama-3.1-8B-Instruct2026.06 | 30.23 | 39.23 | |
| EraRAGBackbone=Llama-3.1-8B-Instruct2026.06 | 26.7 | 38.93 | |
| HiRAGBackbone=Llama-3.1-8B-Instruct2026.06 | 26.2 | 40.7 | |
| HippoRAGBackbone=Llama-3.1-8B-Instruct2026.06 | 20 | 32.54 | |
| H-LightRAGBackbone=Llama-3.1-8B-Instruct2026.06 | 18.73 | 29.41 | |
| G-LightRAGBackbone=Llama-3.1-8B-Instruct2026.06 | 16.01 | 27.73 | |
| L-GraphRAGBackbone=Llama-3.1-8B-Instruct2026.06 | 14.57 | 28.72 | |
| ArchRAGBackbone=Llama-3.1-8B-Instruct2026.06 | 14.27 | 29.22 | |
| Zero-shotBackbone=Llama-3.1-8B-Instruct2026.06 | 8.76 | 19.19 | |
| CoTBackbone=Llama-3.1-8B-Instruct2026.06 | 8.53 | 18.95 |