Mathematical Reasoning on MATH 500 (M500 Score)
95.8M500 ScoreGEAR
Evaluation results
| Method | Links | |
|---|---|---|
| GEARBackbone=Qwen3-4B, Evaluation Context=Standard2026.05 | 95.8 | |
| GEARBackbone=Qwen3-4B, Evaluation Context=Multi-Domain Generalization2026.05 | 95.2 | |
| OPSD+RLBackbone=Qwen3-4B, Evaluation Context=Standard2026.05 | 94.9 | |
| GEARBackbone=Qwen3-8B, Evaluation Context=Standard2026.05 | 94.7 | |
| OPSDBackbone=Qwen3-4B, Evaluation Context=Standard2026.05 | 93.7 | |
| GRPOBackbone=Qwen3-4B, Evaluation Context=Multi-Domain Generalization2026.05 | 93.3 | |
| ARPOBackbone=Qwen3-4B, Evaluation Context=Standard2026.05 | 93.2 | |
| GRPOBackbone=Qwen3-8B, Evaluation Context=Standard2026.05 | 93.2 | |
| MT-GRPOBackbone=Qwen3-4B, Evaluation Context=Standard2026.05 | 93.1 | |
| OPSDBackbone=Qwen3-8B, Evaluation Context=Standard2026.05 | 93.1 | |
| ARPOBackbone=Qwen3-8B, Evaluation Context=Standard2026.05 | 92.8 | |
| MT-GRPOBackbone=Qwen3-8B, Evaluation Context=Standard2026.05 | 92.7 | |
| OPSD+RLBackbone=Qwen3-8B, Evaluation Context=Standard2026.05 | 92.5 | |
| GRPOBackbone=Qwen3-4B, Evaluation Context=Standard2026.05 | 92.4 | |
| BaseBackbone=Qwen3-8B, Evaluation Context=Standard2026.05 | 89.6 | |
| BaseBackbone=Qwen3-4B, Evaluation Context=Standard2026.05 | 88.1 |