Multi-Hop Question Answering on BenchEE Mismatch (Human evaluation score)
90.9Human Evaluation ScoreMDER-DR
Evaluation results
| Method | Links | |
|---|---|---|
| MDER-DRArchitecture=MDER-DR2026.03 | 90.9 | |
| GraphRAGArchitecture=GraphRAG2026.03 | 73.4 | |
| LLM-onlyArchitecture=LLM-only2026.03 | 63.6 | |
| Vector-RAGArchitecture=Vector-RAG2026.03 | 62.4 |