Multi-Hop Question Answering on MuSiQue (test) (Pass@3)
31.1Pass@3PivoARL
Evaluation Results
| Method | Links | |
|---|---|---|
| PivoARLBase Model=Qwen3-4B, Search-augmented=true, Domain=Out-of-domain2026.07 | 31.1 | |
| MetaRLBase Model=Qwen3-4B, Search-augmented=true, Domain=Out-of-domain2026.07 | 24.4 | |
| ZeroSearchBase Model=Qwen3-4B, Search-augmented=true, Domain=Out-of-domain2026.07 | 23.1 | |
| Search-R1Base Model=Qwen3-4B, Search-augmented=true, Domain=Out-of-domain2026.07 | 22.9 | |
| StepSearchBase Model=Qwen3-4B, Search-augmented=true, Domain=Out-of-domain2026.07 | 22.4 | |
| GRPOBase Model=Qwen3-4B, Search-augmented=true, Domain=Out-of-domain2026.07 | 20.7 | |
| GiGPOBase Model=Qwen3-4B, Search-augmented=true, Domain=Out-of-domain2026.07 | 20 | |
| GSPOBase Model=Qwen3-4B, Search-augmented=true, Domain=Out-of-domain2026.07 | 18.5 |