Mathematical Reasoning on SVAMP (Pass@5, # Tokens)
97Pass@50-CoT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| 0-CoTModel=Qwen 3 4B2025.11 | 97 | 1,871 | |
| CoReModel=Qwen 3 4B2025.11 | 97 | 1,844 | |
| 0-CoTModel=Qwen 3 14B2025.11 | 97 | 1,487 | |
| PSModel=Qwen 3 14B2025.11 | 97 | 1,457 | |
| 0-CoTModel=Qwen 3 32B2025.11 | 97 | 1,377 | |
| PSModel=Qwen 3 4B2025.11 | 96.67 | 1,706 | |
| CoReModel=Qwen 3 14B2025.11 | 96.67 | 1,481 | |
| CoReModel=Qwen 3 0.6B2025.11 | 96.33 | 1,380 | |
| CoReModel=Qwen 3 32B2025.11 | 96.33 | 1,464 | |
| PSModel=Qwen 3 0.6B2025.11 | 96 | 1,397 | |
| PSModel=Qwen 3 32B2025.11 | 96 | 1,395 | |
| 0-CoTModel=Qwen 3 0.6B2025.11 | 95.67 | 1,358 | |
| F-CoTModel=Qwen 3 14B2025.11 | 95.67 | 476 | |
| F-CoTModel=Qwen 3 32B2025.11 | 95.67 | 429 | |
| F-CoTModel=Qwen 3 4B2025.11 | 95 | 535 | |
| F-CoTModel=Qwen 3 0.6B2025.11 | 94.33 | 434 |