Question Answering on Average (PopQA, NQ, HotpotQA, 2Wiki, MuSiQue)
0.519EMRouteRAG
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| RouteRAGBackbone=Qwen2.5-7B2025.12 | 0.519 | 0.606 | |
| Search-R1Backbone=Qwen2.5-7B2025.12 | 0.486 | 0.568 | |
| HippoRAG 2Backbone=GPT-4o-mini2025.12 | 0.474 | 0.612 | |
| RouteRAGBackbone=Qwen2.5-3B2025.12 | 0.47 | 0.557 | |
| HippoRAGBackbone=GPT-4o-mini2025.12 | 0.419 | 0.544 | |
| Search-R1Backbone=Qwen2.5-3B2025.12 | 0.404 | 0.493 | |
| RAPTORBackbone=GPT-4o-mini2025.12 | 0.395 | 0.524 | |
| GraphRAGBackbone=GPT-4o-mini2025.12 | 0.386 | 0.555 | |
| R1-SearcherBackbone=Qwen2.5-7B2025.12 | 0.375 | 0.473 | |
| Direct InferenceBackbone=GPT-4o-mini2025.12 | 0.243 | 0.349 | |
| HippoRAG 2Backbone=Qwen2.5-3B2025.12 | 0.229 | 0.345 | |
| Vanilla RAGBackbone=Qwen2.5-3B2025.12 | 0.216 | 0.32 | |
| Vanilla RAGBackbone=Qwen2.5-7B2025.12 | 0.19 | 0.344 | |
| HippoRAG 2Backbone=Qwen2.5-7B2025.12 | 0.183 | 0.34 | |
| Search-o1Backbone=Qwen2.5-3B2025.12 | 0.153 | 0.221 | |
| Search-o1Backbone=Qwen2.5-7B2025.12 | 0.091 | 0.139 | |
| LightRAGBackbone=GPT-4o-mini2025.12 | 0.038 | 0.144 |