Question Answering on Loong
81.75Accuracy (Set 1)PluriHopRAG
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| PluriHopRAGBase LLM=GPT-4o2025.10 | 81.75 | 75.2 | 68.23 | 56.48 | 68.53 | |
| Long-contextBase LLM=GPT-4o2025.10 | 70.4 | 58.38 | 46.95 | 31.11 | 51.71 | |
| StructRAGBase LLM=Qwen2-72B2025.10 | 69.43 | 60.95 | 57.92 | 51.42 | 59.93 | |
| RQ-RAGBase LLM=Qwen2-72B2025.10 | 53.51 | 47.09 | 40.93 | 31.91 | 43.36 | |
| NaiveRAGBase LLM=GPT-4o2025.10 | 50.55 | 49.96 | 45.99 | 33.82 | 45.08 | |
| GraphRAGBase LLM=Qwen2-72B2025.10 | 40.82 | 33.06 | 33.28 | 23.47 | 32.66 |