Code Generation on MBPP (Avg@8)
82.29Avg@8CodeScaler
Evaluation results
| Method | Links | |
|---|---|---|
| CodeScalerTraining Dataset=rStarCoder, Backbone Model=Qwen3-14B, Training Protocol=CodeScaler2026.02 | 82.29 | |
| CodeScalerTraining Dataset=KodCode, Backbone Model=Qwen3-14B, Training Protocol=CodeScaler2026.02 | 81.9 | |
| RLVR (Binary)Training Dataset=KodCode, Backbone Model=Qwen3-14B, Training Protocol=RLVR2026.02 | 78.39 | |
| RLVR (Binary)Training Dataset=rStarCoder, Backbone Model=Qwen3-14B, Training Protocol=RLVR2026.02 | 77.54 | |
| CodeScalerTraining Dataset=KodCode, Backbone Model=Qwen3-8B, Training Protocol=CodeScaler2026.02 | 76.41 | |
| RLVR (Binary)Training Dataset=rStarCoder, Backbone Model=Qwen3-8B, Training Protocol=RLVR2026.02 | 75.79 | |
| RLVR (Binary)Training Dataset=KodCode, Backbone Model=Qwen3-8B, Training Protocol=RLVR2026.02 | 73.24 | |
| CodeScalerTraining Dataset=rStarCoder, Backbone Model=Qwen3-8B, Training Protocol=CodeScaler2026.02 | 73.07 | |
| Qwen3-14B-BaseTraining Dataset=KodCode, Backbone Model=Qwen3-14B, Training Protocol=Base2026.02 | 71.09 | |
| Qwen3-14B-BaseTraining Dataset=rStarCoder, Backbone Model=Qwen3-14B, Training Protocol=Base2026.02 | 71.09 | |
| Qwen3-8B-BaseTraining Dataset=KodCode, Backbone Model=Qwen3-8B, Training Protocol=Base2026.02 | 61.7 | |
| Qwen3-8B-BaseTraining Dataset=rStarCoder, Backbone Model=Qwen3-8B, Training Protocol=Base2026.02 | 61.7 |