Multi-Hop Question Answering on WikiQA
80ScoreMDER-DR
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MDER-DREvaluator=Soft EM, Match=true2026.03 | 80 | 49 | |
| MDER-DREvaluator=LLM-as-a-Judge, Match=true2026.03 | 79 | 66 | |
| MDER-DREvaluator=Soft EM, Match=false2026.03 | 69.9 | 30 | |
| MDER-DREvaluator=LLM-as-a-Judge, Match=false2026.03 | 69.8 | 47 | |
| Vector-RAGEvaluator=Soft EM, Match=true2026.03 | 53.8 | — | |
| Vector-RAGEvaluator=Soft EM, Match=false2026.03 | 53.8 | — | |
| GraphRAGEvaluator=Soft EM, Match=false2026.03 | 51.6 | — | |
| LLM-onlyEvaluator=Soft EM, Match=true2026.03 | 51 | — | |
| GraphRAGEvaluator=Soft EM, Match=true2026.03 | 51 | — | |
| Vector-RAGEvaluator=LLM-as-a-Judge, Match=true2026.03 | 47.7 | — | |
| Vector-RAGEvaluator=LLM-as-a-Judge, Match=false2026.03 | 47.6 | — | |
| GraphRAGEvaluator=LLM-as-a-Judge, Match=false2026.03 | 42.4 | — | |
| GraphRAGEvaluator=LLM-as-a-Judge, Match=true2026.03 | 42.2 | — | |
| LLM-onlyEvaluator=LLM-as-a-Judge, Match=true2026.03 | 41.2 | — | |
| LLM-onlyEvaluator=Soft EM, Match=false2026.03 | 40.8 | — | |
| LLM-onlyEvaluator=LLM-as-a-Judge, Match=false2026.03 | 36.3 | — |