Multi-hop QA on Bamboogle
53.61F1 ScoreReSearch
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| ReSearch2026.02 | 53.61 | 40.8 | 0.41 | 0.24 | |
| GRPOmemory=none2026.02 | 53.18 | 40.8 | 0.79 | 0.28 | |
| MemPO2026.02 | 52.9 | 36.8 | 0.29 | 0.14 | |
| DeepResearcher2026.02 | 48.48 | 37.6 | 1.22 | 0.3 | |
| qwen2.5mode=ReAct2026.02 | 43.7 | 32.8 | 0.95 | 0.24 |