Math Reasoning on AIME 24 (Pass@1, Pass@16)
58.65Pass@1 ScoreGSPO + NSR
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GSPO + NSRModel=Qwen3-30B-A3B-Base, Setting=MoE + 20k, Zero-shot=true, k=32, Temperature=1.02026.05 | 58.65 | 83.26 | |
| GSPOModel=Qwen3-30B-A3B-Base, Setting=MoE + 20k, Zero-shot=true, k=32, Temperature=1.02026.05 | 54.17 | 79.74 | |
| DAPO + NSRModel=Qwen3-8B-Base, Setting=Dense + 30k, Zero-shot=true, k=32, Temperature=1.02026.05 | 43.65 | 75.39 | |
| DAPO + NSRModel=Qwen2.5-Math-7B, Setting=Dense + 8k, Zero-shot=true, k=32, Temperature=1.02026.05 | 40.76 | 57.7 | |
| DAPOModel=Qwen3-8B-Base, Setting=Dense + 30k, Zero-shot=true, k=32, Temperature=1.02026.05 | 37.29 | 75.18 | |
| DAPOModel=Qwen2.5-Math-7B, Setting=Dense + 8k, Zero-shot=true, k=32, Temperature=1.02026.05 | 35.83 | 55.05 |