Question Answering on Aging Dam
21.5BLEUHypercube-RAG
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Hypercube-RAGRetrieval Strategy=Our Method, Base Model=GPT-4o2026.05 | 21.5 | 82.6 | 86.9 | 77.8 | |
| e5Retrieval Strategy=Sparse and Dense Embedding Retrieval, Base Model=GPT-4o2026.05 | 19.7 | 79.8 | 88.9 | 75.5 | |
| NV-EmbedRetrieval Strategy=Sparse and Dense Embedding Retrieval, Base Model=GPT-4o2026.05 | 18.8 | 79.6 | 79.4 | 73.6 | |
| BM25Retrieval Strategy=Sparse and Dense Embedding Retrieval, Base Model=GPT-4o2026.05 | 18.1 | 80.2 | 85.7 | 75.9 | |
| ContrieverRetrieval Strategy=Sparse and Dense Embedding Retrieval, Base Model=GPT-4o2026.05 | 16.7 | 76.8 | 70.9 | 60.9 | |
| LightRAGRetrieval Strategy=Graph-based RAG, Base Model=GPT-4o2026.05 | 12.8 | 75.4 | 65.9 | 46.5 | |
| HippoRAGRetrieval Strategy=Graph-based RAG, Base Model=GPT-4o2026.05 | 12.4 | 73.8 | 80.9 | 58.6 | |
| GraphRAGRetrieval Strategy=Graph-based RAG, Base Model=GPT-4o2026.05 | 11.8 | 72.4 | 72.5 | 58.6 | |
| HippoRAG 2Retrieval Strategy=Graph-based RAG, Base Model=GPT-4o2026.05 | 11.5 | 74.1 | 88.2 | 53.7 | |
| DeepSeek-R1Retrieval Strategy=No Retrieval2026.05 | 8.7 | 71.6 | 35.4 | 18.8 | |
| GPT-4oRetrieval Strategy=No Retrieval2026.05 | 5.5 | 62.5 | 56.9 | 19.6 | |
| Llama-4-Scout-17B-16E-InstructRetrieval Strategy=No Retrieval2026.05 | 5.2 | 64.2 | 34.4 | 15.1 | |
| Llama-3.3-70B-InstructRetrieval Strategy=No Retrieval2026.05 | 2.4 | 46.7 | 39.7 | 17.1 | |
| Qwen2.5-7B-InstructRetrieval Strategy=No Retrieval2026.05 | 1.9 | 43.4 | 15.5 | 6.5 |