Mathematical Reasoning on Olympiad (Acc@16)
40.5Acc@16Mixed-Train
Evaluation Results
| Method | Links | |
|---|---|---|
| Mixed-TrainBackbone=Qwen2.5-7B-Instruct2026.02 | 40.5 | |
| Verify-InitBackbone=Qwen2.5-7B-Instruct2026.02 | 39 | |
| Verify-AlterBackbone=Qwen2.5-7B-Instruct2026.02 | 39 | |
| GenerateBackbone=Qwen2.5-7B-Instruct2026.02 | 37.8 | |
| Verify-AlterBackbone=Qwen2.5-3B-Instruct2026.02 | 30.2 | |
| Verify-InitBackbone=Qwen2.5-3B-Instruct2026.02 | 29.5 | |
| Mixed-TrainBackbone=Qwen2.5-3B-Instruct2026.02 | 27.5 | |
| GenerateBackbone=Qwen2.5-3B-Instruct2026.02 | 27.4 | |
| Verify-AlterBackbone=Qwen2.5-1.5B-Instruct2026.02 | 22.4 | |
| Verify-InitBackbone=Qwen2.5-1.5B-Instruct2026.02 | 21.9 | |
| Mixed-TrainBackbone=Qwen2.5-1.5B-Instruct2026.02 | 21.2 | |
| GenerateBackbone=Qwen2.5-1.5B-Instruct2026.02 | 20.6 |