Mathematical Reasoning on AMC23 (Response Accuracy %)
90.5Response AccuracyS-GRPO
Evaluation Results
| Method | Links | |
|---|---|---|
| S-GRPOModel Scale=7B2026.06 | 90.5 | |
| SuCoModel Scale=7B2026.06 | 90.3 | |
| LHRMsModel Scale=7B2026.06 | 87.3 | |
| AdaptThinkModel Scale=7B2026.06 | 82 | |
| AdaCoTModel Scale=7B2026.06 | 81.5 | |
| DeepSeek-R1-DistillModel Scale=7B2026.06 | 75.5 | |
| SuCoModel Scale=1.5B2026.06 | 73.8 | |
| LHRMsModel Scale=1.5B2026.06 | 70 | |
| S-GRPOModel Scale=1.5B2026.06 | 69.5 | |
| AdaptThinkModel Scale=1.5B2026.06 | 65.8 | |
| AdaCoTModel Scale=1.5B2026.06 | 62.5 | |
| DeepSeek-R1-DistillModel Scale=1.5B2026.06 | 56.5 | |
| Math-InstructModel Scale=7B2026.06 | 53.5 | |
| Math-InstructModel Scale=1.5B2026.06 | 43.5 | |
| Math-BaseModel Scale=7B2026.06 | 36.2 | |
| Math-BaseModel Scale=1.5B2026.06 | 23.9 |