Question Answering on Hurricane
13.2BLEUHypercube-RAG
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Hypercube-RAGRetrieval Strategy=Our Method, Base Model=GPT-4o2026.05 | 13.2 | 82.5 | 80.6 | 66.2 | |
| NV-EmbedRetrieval Strategy=Sparse and Dense Embedding Retrieval, Base Model=GPT-4o2026.05 | 12.2 | 79.1 | 76.9 | 59.6 | |
| e5Retrieval Strategy=Sparse and Dense Embedding Retrieval, Base Model=GPT-4o2026.05 | 11.9 | 79.1 | 78.3 | 63.9 | |
| ContrieverRetrieval Strategy=Sparse and Dense Embedding Retrieval, Base Model=GPT-4o2026.05 | 11.3 | 78.6 | 75.7 | 60.6 | |
| BM25Retrieval Strategy=Sparse and Dense Embedding Retrieval, Base Model=GPT-4o2026.05 | 10.5 | 77.5 | 74.3 | 60.8 | |
| HippoRAG 2Retrieval Strategy=Graph-based RAG, Base Model=GPT-4o2026.05 | 8.6 | 76.3 | 86.7 | 53.5 | |
| LightRAGRetrieval Strategy=Graph-based RAG, Base Model=GPT-4o2026.05 | 7.6 | 72.2 | 64.1 | 47.6 | |
| HippoRAGRetrieval Strategy=Graph-based RAG, Base Model=GPT-4o2026.05 | 6.8 | 73.6 | 70.5 | 48.8 | |
| GraphRAGRetrieval Strategy=Graph-based RAG, Base Model=GPT-4o2026.05 | 6.5 | 72.8 | 68.8 | 50.9 | |
| GPT-4oRetrieval Strategy=No Retrieval2026.05 | 2.6 | 66.8 | 57.1 | 23.5 | |
| Llama-4-Scout-17B-16E-InstructRetrieval Strategy=No Retrieval2026.05 | 2.3 | 62.7 | 53.9 | 39.1 | |
| DeepSeek-R1Retrieval Strategy=No Retrieval2026.05 | 2.2 | 69.2 | 53.4 | 28.5 | |
| Llama-3.3-70B-InstructRetrieval Strategy=No Retrieval2026.05 | 2 | 60.2 | 39.8 | 14.6 | |
| Qwen2.5-7B-InstructRetrieval Strategy=No Retrieval2026.05 | 1.2 | 46.9 | 23.5 | 6.9 |