Multi-Hop QA on Bamboogle (pass@1, Avg Rank)
30.4Pass@1SEARL
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SEARL2026.04 | 30.4 | 1.43 | |
| DAPO2026.04 | 24.8 | 3 | |
| ARPO2026.04 | 17.6 | 3.57 | |
| GRPO2026.04 | 16 | 2.43 | |
| TIR Prompt2026.04 | 12 | 5.29 | |
| Reinforce++2026.04 | 0.8 | 4.57 |
| Method | Links | ||
|---|---|---|---|
| SEARL2026.04 | 30.4 | 1.43 | |
| DAPO2026.04 | 24.8 | 3 | |
| ARPO2026.04 | 17.6 | 3.57 | |
| GRPO2026.04 | 16 | 2.43 | |
| TIR Prompt2026.04 | 12 | 5.29 | |
| Reinforce++2026.04 | 0.8 | 4.57 |