Multi-Hop QA on 2wiki (pass@1, Avg Rank)
36pass@1SEARL
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SEARL2026.04 | 36 | 1.43 | |
| DAPO2026.04 | 35 | 3 | |
| GRPO2026.04 | 34.5 | 2.43 | |
| Reinforce++2026.04 | 26 | 4.57 | |
| ARPO2026.04 | 22 | 3.57 | |
| TIR Prompt2026.04 | 12.5 | 5.29 |
| Method | Links | ||
|---|---|---|---|
| SEARL2026.04 | 36 | 1.43 | |
| DAPO2026.04 | 35 | 3 | |
| GRPO2026.04 | 34.5 | 2.43 | |
| Reinforce++2026.04 | 26 | 4.57 | |
| ARPO2026.04 | 22 | 3.57 | |
| TIR Prompt2026.04 | 12.5 | 5.29 |