Multi-hop Question Answering on HotpotQA Large-scale corpus (KILT)
53.4Exact Match (EM)Agentic Hybrid
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Agentic HybridCorpus=Large-scale (KILT)2026.05 | 53.4 | 66.8 | 79.4 | |
| LOGICALRAGCorpus=Large-scale (KILT)2026.05 | 52.4 | 66.4 | 78.4 | |
| MA-RAGCorpus=Large-scale (KILT)2026.05 | 33.7 | 45.7 | 52.3 |