Multi-hop Question Answering on 2WikiMultiHopQA (Str-Acc, LLM-Acc)
70.3String AccuracyMemGraphRAG
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MemGraphRAGRetrieval Strategy=Graph-based Retrieval-Augmented-Generation2026.05 | 70.3 | 69.8 | |
| ConRAGLLM backbone=Gemma-4-31B, Type=Ours2026.05 | 70.2 | 77.6 | |
| LinearRAGRetrieval Strategy=Graph-based Retrieval-Augmented-Generation2026.05 | 70.2 | 65.7 | |
| Youtu-GraphRAGLLM backbone=Gemma-4-31B, Type=Graph-based RAG2026.05 | 69.5 | 74.7 | |
| GFM-RAGRetrieval Strategy=Graph-based Retrieval-Augmented-Generation2026.05 | 69.1 | 61.1 | |
| HippoRAGRetrieval Strategy=Graph-based Retrieval-Augmented-Generation2026.05 | 67.5 | 61.3 | |
| ConRAGLLM backbone=GPT-4o-mini, Type=Ours2026.05 | 64.9 | 70.7 | |
| LogicRAGRetrieval Strategy=Graph-based Retrieval-Augmented-Generation2026.05 | 64.8 | 63.4 | |
| LogicRAGLLM backbone=GPT-4o-mini, Type=Reasoning-based RAG2026.05 | 64.7 | 62.5 | |
| HippoRAG2Retrieval Strategy=Graph-based Retrieval-Augmented-Generation2026.05 | 64.2 | 57.9 | |
| LogicRAGLLM backbone=Gemma-4-31B, Type=Reasoning-based RAG2026.05 | 60.7 | 65.7 | |
| E^2GraphRAGRetrieval Strategy=Graph-based Retrieval-Augmented-Generation2026.05 | 57.2 | 40.9 | |
| LightRAGRetrieval Strategy=Graph-based Retrieval-Augmented-Generation2026.05 | 56.9 | 40.5 | |
| HippoRAGLLM backbone=Gemma-4-31B, Type=Graph-based RAG2026.05 | 52.9 | 55.6 | |
| RAPTORRetrieval Strategy=Graph-based Retrieval-Augmented-Generation2026.05 | 51.7 | 43.6 | |
| HippoRAG2LLM backbone=GPT-4o-mini, Type=Graph-based RAG2026.05 | 50 | 47.1 | |
| RetrievalRetrieval Strategy=Vanilla Retrieval-Augmented-Generation, Top-k Retrieval=52026.05 | 49.8 | 45.4 | |
| G-retrieverRetrieval Strategy=Graph-based Retrieval-Augmented-Generation2026.05 | 47.8 | 29.7 | |
| RAPTORLLM backbone=GPT-4o-mini, Type=Graph-based RAG2026.05 | 47.7 | 45.9 | |
| HippoRAGLLM backbone=GPT-4o-mini, Type=Graph-based RAG2026.05 | 47.7 | 47.2 | |
| KGPLLM backbone=GPT-4o-mini, Type=Graph-based RAG2026.05 | 47.5 | 43.7 | |
| MS-GraphRAGRetrieval Strategy=Graph-based Retrieval-Augmented-Generation2026.05 | 47.3 | 38.6 | |
| LazyGraphRAGRetrieval Strategy=Graph-based Retrieval-Augmented-Generation2026.05 | 46.8 | 37.9 | |
| Vanilla RAG (Top-5)LLM backbone=GPT-4o-mini, Type=Text RAG2026.05 | 46.7 | 45.6 | |
| RetrievalRetrieval Strategy=Vanilla Retrieval-Augmented-Generation, Top-k Retrieval=32026.05 | 46.5 | 41.8 | |
| GraphRAGLLM backbone=GPT-4o-mini, Type=Graph-based RAG2026.05 | 46.3 | 43.3 | |
| HippoRAG2LLM backbone=Gemma-4-31B, Type=Graph-based RAG2026.05 | 46.1 | 47.5 | |
| LinearRAGLLM backbone=GPT-4o-mini, Type=Graph-based RAG2026.05 | 44.7 | 45.7 | |
| LightRAGLLM backbone=GPT-4o-mini, Type=Graph-based RAG2026.05 | 43.1 | 36.3 | |
| Vanilla RAG (Top-3)LLM backbone=GPT-4o-mini, Type=Text RAG2026.05 | 43 | 42 | |
| LinearRAGLLM backbone=Gemma-4-31B, Type=Graph-based RAG2026.05 | 43 | 44.2 | |
| Vanilla RAG (Top-5)LLM backbone=Gemma-4-31B, Type=Text RAG2026.05 | 40.3 | 40.3 | |
| RetrievalRetrieval Strategy=Vanilla Retrieval-Augmented-Generation, Top-k Retrieval=12026.05 | 38.1 | 34 | |
| Vanilla RAG (Top-3)LLM backbone=Gemma-4-31B, Type=Text RAG2026.05 | 37.5 | 36.9 | |
| GPT-4o-miniRetrieval Strategy=Direct Zero-shot LLM Inference2026.05 | 36 | 31.3 | |
| Vanilla RAG (Top-1)LLM backbone=GPT-4o-mini, Type=Text RAG2026.05 | 34.8 | 37.3 | |
| Llama3-8BRetrieval Strategy=Direct Zero-shot LLM Inference2026.05 | 34 | 16 | |
| KGPRetrieval Strategy=Graph-based Retrieval-Augmented-Generation2026.05 | 33.1 | 32.7 | |
| Zero-shotLLM backbone=Gemma-4-31B, Type=Direct LLM2026.05 | 31.8 | 31.4 | |
| GPT-3.5-TurboRetrieval Strategy=Direct Zero-shot LLM Inference2026.05 | 28.3 | 31 | |
| G-RetrieverLLM backbone=GPT-4o-mini, Type=Graph-based RAG2026.05 | 26.7 | 35.7 | |
| Zero-shotLLM backbone=GPT-4o-mini, Type=Direct LLM2026.05 | 26.4 | 24.3 | |
| Llama3-13BRetrieval Strategy=Direct Zero-shot LLM Inference2026.05 | 22.3 | 9.5 | |
| Vanilla RAG (Top-1)LLM backbone=Gemma-4-31B, Type=Text RAG2026.05 | 14.3 | 13.3 |