Multi-hop Question Answering on HotpotQA N=60 (test)
1Average Roundsfixed_k1
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| fixed_k1Policy=fixed_k12026.06 | 1 | 86 | 0.03 | 0.17 | — | |
| shp (full)Policy=shp (full)2026.06 | 2.4 | 129 | 0.004 | 0.78 | — | |
| oracle_isPolicy=oracle_is2026.06 | 2.73 | 170 | 0.115 | 3 | — | |
| fixed_k3Policy=fixed_k32026.06 | 3 | 53 | 0.001 | 0.97 | — | |
| entropy_onlyPolicy=entropy_only2026.06 | 3.92 | 38 | 0.004 | 0.81 | — | |
| fixed_k6 (base)Policy=fixed_k62026.06 | 6 | — | — | — | — | |
| critic_onlyPolicy=critic_only2026.06 | 6 | 0 | 0 | — | — |