Mathematical Reasoning on HMMT25 (mean@8)
13.33Mean Score (8 samples)ACPO
Evaluation Results
| Method | Links | |
|---|---|---|
| ACPOModel=DeepSeek-R1-Distill-Qwen-7B2026.07 | 13.33 | |
| GTPOModel=DeepSeek-R1-Distill-Qwen-7B2026.07 | 13.19 | |
| SAPOModel=DeepSeek-R1-Distill-Qwen-7B2026.07 | 12.46 | |
| ACPOModel=Qwen3-8B-Base2026.07 | 11.04 | |
| 80/20Model=DeepSeek-R1-Distill-Qwen-7B2026.07 | 10.83 | |
| SAPOModel=Qwen3-8B-Base2026.07 | 10 | |
| ACPOModel=Qwen2.5-Math-7B2026.07 | 9.86 | |
| DAPOModel=DeepSeek-R1-Distill-Qwen-7B2026.07 | 9.61 | |
| GTPOModel=Qwen3-8B-Base2026.07 | 8.83 | |
| GRPOModel=DeepSeek-R1-Distill-Qwen-7B2026.07 | 8.75 | |
| GTPOModel=Qwen2.5-Math-7B2026.07 | 8.7 | |
| SAPOModel=Qwen2.5-Math-7B2026.07 | 8.64 | |
| 80/20Model=Qwen3-8B-Base2026.07 | 7.91 | |
| 80/20Model=Qwen2.5-Math-7B2026.07 | 7.7 | |
| DAPOModel=Qwen3-8B-Base2026.07 | 7.12 | |
| DAPOModel=Qwen2.5-Math-7B2026.07 | 6.67 | |
| GRPOModel=Qwen3-8B-Base2026.07 | 6.67 | |
| GRPOModel=Qwen2.5-Math-7B2026.07 | 5.2 |