Question Answering on HotpotQA (Exact Match Accuracy)
42.9Exact Match Accuracy (EM)ParallelSearch
Evaluation Results
| Method | Links | |
|---|---|---|
| ParallelSearchModel=Qwen2.5-7B2026.03 | 42.9 | |
| ICRLModel=Qwen2.5-7B2026.03 | 42.6 | |
| Search-R1Model=Qwen2.5-7B2026.03 | 37 | |
| ICRLModel=Qwen2.5-3B2026.03 | 35.4 | |
| ZeroSearchModel=Qwen2.5-7B2026.03 | 34.6 | |
| Reject SamplingModel=Qwen2.5-7B2026.03 | 33.1 | |
| Search-R1Model=Qwen2.5-3B2026.03 | 32.4 | |
| RAGModel=Qwen2.5-7B2026.03 | 29.9 | |
| ZeroSearchModel=Qwen2.5-3B2026.03 | 27.4 | |
| RAGModel=Qwen2.5-3B2026.03 | 25.5 | |
| R1-baseModel=Qwen2.5-7B2026.03 | 24.2 | |
| Reject SamplingModel=Qwen2.5-3B2026.03 | 24 | |
| R1-instructModel=Qwen2.5-7B2026.03 | 23.7 | |
| Search-o1Model=Qwen2.5-3B2026.03 | 22.1 | |
| SFTModel=Qwen2.5-7B2026.03 | 21.7 | |
| R1-instructModel=Qwen2.5-3B2026.03 | 20.8 | |
| Search-o1Model=Qwen2.5-7B2026.03 | 18.7 | |
| SFTModel=Qwen2.5-3B2026.03 | 18.6 | |
| DirectModel=Qwen2.5-7B2026.03 | 18.3 | |
| IRCoTModel=Qwen2.5-3B2026.03 | 16.4 | |
| DirectModel=Qwen2.5-3B2026.03 | 14.9 | |
| IRCoTModel=Qwen2.5-7B2026.03 | 13.3 | |
| CoTModel=Qwen2.5-7B2026.03 | 9.2 | |
| CoTModel=Qwen2.5-3B2026.03 | 2.1 |