Question Answering on MetaQA 1-hop (Acc, F1)
82.77AccuracyGraph-S^3
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Graph-S^3Generator=Finetuned-7B2025.10 | 82.77 | 92.04 | |
| Graph-S^3Generator=Qwen2-7B2025.10 | 81.5 | 90.22 | |
| KG-AgentGenerator=Finetuned-7B2025.10 | 75.99 | 86.02 | |
| RAG/1hopGenerator=Qwen2-7B2025.10 | 75.93 | 86.36 | |
| KG-AgentGenerator=Qwen2-7B2025.10 | 75.8 | 85.92 | |
| Graph-S^3Generator=Llama3-8B2025.10 | 67.5 | 76.56 | |
| KG-AgentGenerator=Llama3-8B2025.10 | 65.55 | 75.03 | |
| RAG/1hopGenerator=Llama3-8B2025.10 | 60.17 | 70.84 | |
| RAG/1hopGenerator=Finetuned-7B2025.10 | 59.5 | 69.54 | |
| RAG/2hopGenerator=Qwen2-7B2025.10 | 42.03 | 55.59 | |
| RAG/2hopGenerator=Finetuned-7B2025.10 | 35.83 | 52.03 | |
| RAG/2hopGenerator=Llama3-8B2025.10 | 29.07 | 42.47 | |
| LightRAGGenerator=Finetuned-7B2025.10 | 13.77 | 20.6 | |
| LightRAGGenerator=Llama3-8B2025.10 | 13.13 | 21.47 | |
| ToGGenerator=Llama3-8B2025.10 | 12.4 | 15.88 | |
| No graphGenerator=Llama3-8B2025.10 | 12.2 | 17.6 | |
| G-retrieverGenerator=Finetuned-7B2025.10 | 8.93 | 11.51 | |
| ToGGenerator=Finetuned-7B2025.10 | 2.23 | 3.44 | |
| No graphGenerator=Qwen2-7B2025.10 | 2 | 2.88 | |
| No graphGenerator=Finetuned-7B2025.10 | 1.63 | 2.49 | |
| ToGGenerator=Qwen2-7B2025.10 | 1.37 | 1.75 | |
| LightRAGGenerator=Qwen2-7B2025.10 | 1.13 | 1.76 | |
| G-retrieverGenerator=Llama3-8B2025.10 | 0.67 | 1.56 | |
| G-retrieverGenerator=Qwen2-7B2025.10 | 0.63 | 1.6 | |
| RAG/3hopGenerator=Finetuned-7B2025.10 | 0.57 | 2.99 | |
| RAG/3hopGenerator=Qwen2-7B2025.10 | 0.37 | 3.03 | |
| RAG/3hopGenerator=Llama3-8B2025.10 | 0.33 | 3.67 | |
| No retrieverGenerator=Qwen2-7B2025.10 | 0 | 0.29 | |
| No retrieverGenerator=Llama3-8B2025.10 | 0 | 0.58 | |
| No retrieverGenerator=Finetuned-7B2025.10 | 0 | 0.29 |