Question Answering on HotpotQA (test) (EM/F1)
68.75EMGraph-R1 + EKA
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Graph-R1 + EKABackbone=Qwen2.5-14B-Instruct2025.12 | 68.75 | 74.47 | |
| StandardRAGBackbone=GPT-4o-mini2025.12 | 35.16 | 46.7 | |
| GraphRAGBackbone=GPT-4o-mini2025.12 | 19.53 | 31.67 | |
| NaiveGenerationBackbone=GPT-4o-mini2025.12 | 18.75 | 31.79 | |
| LightRAGBackbone=GPT-4o-mini2025.12 | 18.75 | 30.7 |