Mathematical Reasoning on Math500 (Acc@16)
74Acc@16Verify-Init
Evaluation Results
| Method | Links | |
|---|---|---|
| Verify-InitBackbone=Qwen2.5-7B-Instruct2026.02 | 74 | |
| Mixed-TrainBackbone=Qwen2.5-7B-Instruct2026.02 | 73.5 | |
| Verify-AlterBackbone=Qwen2.5-7B-Instruct2026.02 | 72.9 | |
| GenerateBackbone=Qwen2.5-7B-Instruct2026.02 | 70.9 | |
| Verify-AlterBackbone=Qwen2.5-3B-Instruct2026.02 | 64.5 | |
| Verify-InitBackbone=Qwen2.5-3B-Instruct2026.02 | 63.1 | |
| GenerateBackbone=Qwen2.5-3B-Instruct2026.02 | 59.6 | |
| Mixed-TrainBackbone=Qwen2.5-3B-Instruct2026.02 | 59.2 | |
| Verify-InitBackbone=Qwen2.5-1.5B-Instruct2026.02 | 54.7 | |
| Verify-AlterBackbone=Qwen2.5-1.5B-Instruct2026.02 | 54.2 | |
| Mixed-TrainBackbone=Qwen2.5-1.5B-Instruct2026.02 | 53.8 | |
| GenerateBackbone=Qwen2.5-1.5B-Instruct2026.02 | 53.7 |