General Question Answering on TriviaQA (test val)
68.2EMDeepControl
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DeepControlBackbone Model=Qwen2.5-7B2026.02 | 68.2 | 47.9 | |
| DeepControlBackbone Model=Qwen2.5-3B2026.02 | 64.5 | 41.1 | |
| Search-R1-baseBackbone Model=Qwen2.5-7B2026.02 | 63.8 | 43.1 | |
| Search-R1-instructBackbone Model=Qwen2.5-7B2026.02 | 61 | 38.5 | |
| Rejection SamplingBackbone Model=Qwen2.5-7B2026.02 | 59.2 | 34.8 | |
| Search-R1-baseBackbone Model=Qwen2.5-3B2026.02 | 58.7 | 30.3 | |
| RAGBackbone Model=Qwen2.5-7B2026.02 | 58.5 | 30.4 | |
| Search-R1-instructBackbone Model=Qwen2.5-3B2026.02 | 54.5 | 32.5 | |
| RAGBackbone Model=Qwen2.5-3B2026.02 | 54.4 | 27 | |
| R1-baseBackbone Model=Qwen2.5-7B2026.02 | 53.9 | 27.6 | |
| R1-instructBackbone Model=Qwen2.5-7B2026.02 | 53.7 | 27.1 | |
| Rejection SamplingBackbone Model=Qwen2.5-3B2026.02 | 48.8 | 26.5 | |
| IRCoTBackbone Model=Qwen2.5-7B2026.02 | 47.8 | 23.9 | |
| Search-o1Backbone Model=Qwen2.5-3B2026.02 | 47.2 | 25.5 | |
| R1-baseBackbone Model=Qwen2.5-3B2026.02 | 45.5 | 22.9 | |
| R1-instructBackbone Model=Qwen2.5-3B2026.02 | 44.9 | 22.4 | |
| Search-o1Backbone Model=Qwen2.5-7B2026.02 | 44.3 | 20.6 | |
| Direct InferenceBackbone Model=Qwen2.5-7B2026.02 | 40.8 | 18.1 | |
| SFTBackbone Model=Qwen2.5-7B2026.02 | 35.4 | 20.7 | |
| IRCoTBackbone Model=Qwen2.5-3B2026.02 | 31.2 | 18.1 | |
| SFTBackbone Model=Qwen2.5-3B2026.02 | 29.2 | 17.6 | |
| Direct InferenceBackbone Model=Qwen2.5-3B2026.02 | 28.8 | 13.4 | |
| CoTBackbone Model=Qwen2.5-7B2026.02 | 18.5 | 10.6 | |
| CoTBackbone Model=Qwen2.5-3B2026.02 | 3.2 | 1.5 |