Reasoning on AIME 2025 (Pass@1, Cons@6, Cons@32)
58.8Pass@1SSFT-32B-GFPO
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| SSFT-32B-GFPOTraining data type=multi-target data2025.10 | 58.8 | 78.48 | 83.33 | |
| SSFT-32BTraining data type=multi-target data2025.10 | 58.13 | 73.94 | 86.67 | |
| SSFT-32B (random σ)Training data type=multi-target data2025.10 | 55.1 | 67.58 | 80 | |
| SFT-mixed-distill-32B-tagsTraining data type=multi-target data2025.10 | 51.96 | 70 | 76.67 | |
| SFT-OSS-distill-32BTraining data type=single-target data2025.10 | 48.75 | 65.45 | 73.33 | |
| Multiverse-32BTraining data type=multi-target data2025.10 | 45.8 | — | — | |
| s1.1-32BTraining data type=single-target data2025.10 | 44.27 | 53.33 | 63.33 | |
| Qwen2.5-32B-InstructTraining data type=single-target data2025.10 | 10.4 | — | — |