Reasoning on MATH 500 (Pass@1, Cons@6, Cons@32)
92.16Pass@1s1.1-32B
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| s1.1-32BTraining data type=single-target data2025.10 | 92.16 | 95.6 | 94.8 | |
| Multiverse-32BTraining data type=multi-target data2025.10 | 91.8 | — | — | |
| SSFT-32BTraining data type=multi-target data2025.10 | 90.02 | 96.47 | 96.8 | |
| SSFT-32B (random σ)Training data type=multi-target data2025.10 | 89.95 | 95.67 | 95.6 | |
| SSFT-32B-GFPOTraining data type=multi-target data2025.10 | 89.9 | 95.73 | 96.8 | |
| SFT-OSS-distill-32BTraining data type=single-target data2025.10 | 89.54 | 95.47 | 96 | |
| SFT-mixed-distill-32B-tagsTraining data type=multi-target data2025.10 | 88.49 | 95.88 | 96.2 | |
| Qwen2.5-32B-InstructTraining data type=single-target data2025.10 | 80.4 | — | — |