Multi-hop Question Answering on HotpotQA (full) (EM, F1)
69.2EMGold Context
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Gold ContextLLM=Q32, #TokAvg.=86.5, Info. Density=1.6e-12026.02 | 69.2 | 82.1 | |
| PLUGMEMEmb=NVE, LLM=Q32, #TokAvg.=81.6, Info. Density=1.4e-12026.02 | 61.4 | 74.1 | |
| HippoRAG2Emb=NVE, LLM=Q32, #TokAvg.=595.1, Info. Density=1.9e-22026.02 | 60 | 73.3 | |
| PropRAGEmb=NVE, LLM=Q32, #TokAvg.=626.1, Info. Density=1.9e-22026.02 | 57.8 | 72.1 | |
| RAPTOREmb=NVE, LLM=Q32, #TokAvg.=806.3, Info. Density=1.1e-22026.02 | 56.7 | 69.7 | |
| GraphRAGEmb=NVE, LLM=L702026.02 | 55.2 | 68.6 | |
| Vanilla RetrievalEmb=NVE, LLM=Q32, #TokAvg.=659.2, Info. Density=1.2e-22026.02 | 51.7 | 62.7 | |
| A-MemEmb=NVE, LLM=Q32, #TokAvg.=695.6, Info. Density=1.2e-22026.02 | 43.8 | 53.6 | |
| No ContextLLM=Q322026.02 | 22.1 | 31 |