Multi-Hop Question Answering on 2Wiki (test) (Pass@3)
64Pass@3PivoARL
Evaluation Results
| Method | Links | |
|---|---|---|
| PivoARLBase Model=Qwen3-4B, Search-augmented=true, Domain=Out-of-domain2026.07 | 64 | |
| MetaRLBase Model=Qwen3-4B, Search-augmented=true, Domain=Out-of-domain2026.07 | 55.2 | |
| Search-R1Base Model=Qwen3-4B, Search-augmented=true, Domain=Out-of-domain2026.07 | 52.4 | |
| StepSearchBase Model=Qwen3-4B, Search-augmented=true, Domain=Out-of-domain2026.07 | 50 | |
| GiGPOBase Model=Qwen3-4B, Search-augmented=true, Domain=Out-of-domain2026.07 | 47.9 | |
| ZeroSearchBase Model=Qwen3-4B, Search-augmented=true, Domain=Out-of-domain2026.07 | 47.3 | |
| GRPOBase Model=Qwen3-4B, Search-augmented=true, Domain=Out-of-domain2026.07 | 46.4 | |
| GSPOBase Model=Qwen3-4B, Search-augmented=true, Domain=Out-of-domain2026.07 | 43 |