Mathematical Reasoning on BeyondAIME (mean@8)
21Mean@8 ScoreACPO
Evaluation Results
| Method | Links | |
|---|---|---|
| ACPOModel=DeepSeek-R1-Distill-Qwen-7B2026.07 | 21 | |
| GTPOModel=DeepSeek-R1-Distill-Qwen-7B2026.07 | 19.12 | |
| SAPOModel=DeepSeek-R1-Distill-Qwen-7B2026.07 | 18.62 | |
| 80/20Model=DeepSeek-R1-Distill-Qwen-7B2026.07 | 15.75 | |
| ACPOModel=Qwen3-8B-Base2026.07 | 15.5 | |
| DAPOModel=DeepSeek-R1-Distill-Qwen-7B2026.07 | 15.12 | |
| GRPOModel=DeepSeek-R1-Distill-Qwen-7B2026.07 | 13.62 | |
| GTPOModel=Qwen3-8B-Base2026.07 | 12.62 | |
| SAPOModel=Qwen3-8B-Base2026.07 | 12.25 | |
| 80/20Model=Qwen3-8B-Base2026.07 | 12 | |
| DAPOModel=Qwen3-8B-Base2026.07 | 11.75 | |
| GRPOModel=Qwen3-8B-Base2026.07 | 9.5 | |
| ACPOModel=Qwen2.5-Math-7B2026.07 | 8 | |
| GTPOModel=Qwen2.5-Math-7B2026.07 | 7.12 | |
| SAPOModel=Qwen2.5-Math-7B2026.07 | 6.75 | |
| 80/20Model=Qwen2.5-Math-7B2026.07 | 5.25 | |
| DAPOModel=Qwen2.5-Math-7B2026.07 | 4.5 | |
| GRPOModel=Qwen2.5-Math-7B2026.07 | 4 |