General QA on TriviaQA
71.9EMIRCoT
Evaluation results
| Method | Links | ||
|---|---|---|---|
| IRCoTBackbone=Qwen2.5-72B-Instruct2026.03 | 71.9 | 84.2 | |
| GraphRAG-RouterSystem=Our Method2026.03 | 67.2 | — | |
| Router-R1System=RL-based Agent2026.03 | 66.3 | — | |
| ZeroSearchBackbone=Qwen2.5-7b-Instruct2025.10 | 65.2 | — | |
| TaSR-RAGBackbone=Qwen2.5-72B-Instruct2026.03 | 64.5 | 69.9 | |
| ReSeekBackbone=Qwen2.5-7b-Instruct2025.10 | 64 | — | |
| GraphSearchSystem=Training-free Agent2026.03 | 62.3 | — | |
| RAPTORSystem=RAG-Based2026.03 | 62 | — | |
| HippoRAGBackbone=Qwen2.5-72B-Instruct2026.03 | 61.9 | 71.9 | |
| Search-o1System=Training-free Agent2026.03 | 61.8 | — | |
| GraphRAGBackbone=Qwen2.5-72B-Instruct2026.03 | 61.7 | 70.9 | |
| Search-R1Backbone=Qwen2.5-7b-Instruct2025.10 | 61 | — | |
| StructRAGBackbone=Qwen2.5-72B-Instruct2026.03 | 60.8 | 71.9 | |
| LinearRAGSystem=RAG-Based2026.03 | 60.8 | — | |
| GraphRAGSystem=RAG-Based2026.03 | 60.4 | — | |
| HippoRAG2System=RAG-Based2026.03 | 60.1 | — | |
| HyperGraphRAGSystem=RAG-Based2026.03 | 58.5 | — | |
| RAGBackbone=Qwen2.5-7b-Instruct2025.10 | 58.5 | — | |
| HyperGraphRAGBackbone=Qwen2.5-72B-Instruct2026.03 | 57.4 | 68.1 | |
| ZeroSearchBackbone=Qwen2.5-3b-Instruct2025.10 | 57.4 | — | |
| Search-R1System=RL-based Agent2026.03 | 56.8 | — | |
| Vanilla RAGSystem=RAG-Based2026.03 | 55.7 | — | |
| ReSeekBackbone=Qwen2.5-3b-Instruct2025.10 | 55.3 | — | |
| CoTSystem=Basic LLM2026.03 | 55.2 | — | |
| TaSR-RAGBackbone=Qwen2.5-7B-Instruct2026.03 | 55.1 | 60.5 | |
| Search-R1Backbone=Qwen2.5-3b-Instruct2025.10 | 54.5 | — | |
| RAGBackbone=Qwen2.5-3b-Instruct2025.10 | 54.4 | — | |
| R1Backbone=Qwen2.5-7b-Instruct2025.10 | 53.7 | — | |
| RAGBackbone=Qwen2.5-72B-Instruct2026.03 | 53.4 | 62.5 | |
| CoTBackbone=Qwen2.5-72B-Instruct2026.03 | 51.6 | 60.4 | |
| StructRAGBackbone=Qwen2.5-7B-Instruct2026.03 | 50 | 60.4 | |
| IRCoTBackbone=Qwen2.5-7B-Instruct2026.03 | 47.8 | 65.9 | |
| Search-o1Backbone=Qwen2.5-3b-Instruct2025.10 | 47.2 | — | |
| Graph-R1System=RL-based Agent2026.03 | 46.4 | — | |
| Direct InferenceBackbone=Qwen2.5-72B-Instruct2026.03 | 45.4 | 53.1 | |
| R1Backbone=Qwen2.5-3b-Instruct2025.10 | 44.9 | — | |
| Search-o1Backbone=Qwen2.5-7b-Instruct2025.10 | 44.3 | — | |
| Direct InferenceBackbone=Qwen2.5-7B-Instruct2026.03 | 40.8 | 49.7 | |
| Direct InferenceBackbone=Qwen2.5-7b-Instruct2025.10 | 40.8 | — | |
| HippoRAGBackbone=Qwen2.5-7B-Instruct2026.03 | 37 | 51 | |
| GraphRAGBackbone=Qwen2.5-7B-Instruct2026.03 | 35.9 | 48 | |
| RAGBackbone=Qwen2.5-7B-Instruct2026.03 | 35.5 | 67.2 | |
| Direct InferSystem=Basic LLM2026.03 | 30.8 | — | |
| Direct InferenceBackbone=Qwen2.5-3b-Instruct2025.10 | 28.8 | — | |
| CoTBackbone=Qwen2.5-7B-Instruct2026.03 | 18.5 | 49.7 | |
| CoTBackbone=Qwen2.5-7b-Instruct2025.10 | 18.5 | — | |
| HypergraphRAGBackbone=Qwen2.5-7B-Instruct2026.03 | 3.8 | 6.2 | |
| CoTBackbone=Qwen2.5-3b-Instruct2025.10 | 3.2 | — |