Code Reasoning on HumanEvalPro
46.34Pass@1ACPO
Evaluation results
| Method | Links | |
|---|---|---|
| ACPOModel=Qwen2.5-Math-7B2026.07 | 46.34 | |
| ACPOModel=DeepSeek-R1-Distill-Qwen-7B2026.07 | 45.51 | |
| SAPOModel=Qwen2.5-Math-7B2026.07 | 43.9 | |
| DAPOModel=Qwen2.5-Math-7B2026.07 | 43.29 | |
| GTPOModel=Qwen2.5-Math-7B2026.07 | 43.29 | |
| GRPOModel=DeepSeek-R1-Distill-Qwen-7B2026.07 | 43.29 | |
| SAPOModel=DeepSeek-R1-Distill-Qwen-7B2026.07 | 43.29 | |
| DAPOModel=DeepSeek-R1-Distill-Qwen-7B2026.07 | 42.68 | |
| 80/20Model=DeepSeek-R1-Distill-Qwen-7B2026.07 | 42.68 | |
| GRPOModel=Qwen2.5-Math-7B2026.07 | 42.59 | |
| 80/20Model=Qwen2.5-Math-7B2026.07 | 41.46 | |
| ACPOModel=Qwen3-8B-Base2026.07 | 41.46 | |
| GTPOModel=DeepSeek-R1-Distill-Qwen-7B2026.07 | 41.46 | |
| SAPOModel=Qwen3-8B-Base2026.07 | 40.85 | |
| DAPOModel=Qwen3-8B-Base2026.07 | 39.06 | |
| GTPOModel=Qwen3-8B-Base2026.07 | 38.49 | |
| 80/20Model=Qwen3-8B-Base2026.07 | 37.21 | |
| GRPOModel=Qwen3-8B-Base2026.07 | 35.38 |