Multi-hop Question Answering on MuSiQue offline Wiki-18 (test val)
24.4EMESearch-base
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ESearch-baseModel Scale=7B2025.10 | 24.4 | 37.1 | |
| ESearch-instructModel Scale=7B2025.10 | 24.1 | 35.8 | |
| ESearch-baseModel Scale=3B2025.10 | 23.6 | 34.5 | |
| ESearch-instructModel Scale=3B2025.10 | 23.2 | 33.9 | |
| StepSearch-instructModel Scale=7B2025.10 | 22.6 | 31.2 | |
| ZeroSearch-instructModel Scale=7B2025.10 | 21.9 | 32 | |
| StepSearch-baseModel Scale=7B2025.10 | 21.6 | 32.4 | |
| Search-R1-baseModel Scale=7B2025.10 | 20.6 | 29 | |
| ZeroSearch-baseModel Scale=7B2025.10 | 20.1 | 28.6 | |
| Research-instructModel Scale=7B2025.10 | 18.4 | 27.1 | |
| StepSearch-baseModel Scale=3B2025.10 | 18.1 | 27.3 | |
| Search-R1-instructModel Scale=7B2025.10 | 18.1 | 26.2 | |
| R-Search-instruct-GRPOModel Scale=7B2025.10 | 17.9 | 26 | |
| StepSearch-instructModel Scale=3B2025.10 | 17.4 | 26.1 | |
| SSRL-instructModel Scale=7B2025.10 | 15.3 | 23.8 | |
| Research-baseModel Scale=7B2025.10 | 14.3 | 23 | |
| R-Search-instruct-PPOModel Scale=7B2025.10 | 13.3 | 20.9 | |
| R-Search-instruct-GRPOModel Scale=3B2025.10 | 13.1 | 20.8 | |
| R-Search-instruct-PPOModel Scale=3B2025.10 | 12.4 | 18.7 | |
| Search-R1-instructModel Scale=3B2025.10 | 12 | 18.8 | |
| ZeroSearch-baseModel Scale=3B2025.10 | 9.6 | 16.4 | |
| SSRL-instructModel Scale=3B2025.10 | 9.3 | 15.6 | |
| ZeroSearch-instructModel Scale=3B2025.10 | 8.8 | 14.5 | |
| Search-R1-baseModel Scale=3B2025.10 | 8.1 | 14.6 |