Reasoning on Aggregate (AIME’24, AIME’25, MATH 500, AMC’23, GPQA Diamond) (test)
14.64AAIPPPO
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| PPPOBackbone=Qwen3-8B2025.12 | 14.64 | 0.2483 | 58.95 | |
| PPPOBackbone=Qwen3-4B2025.12 | 12.36 | 0.2617 | 47.24 | |
| PPPOBackbone=Qwen3-1.7B2025.12 | 10.52 | 0.2919 | 36.05 | |
| DAPO-FTBackbone=Qwen3-8B2025.12 | 9.47 | 0.2 | 47.36 | |
| DAPOBackbone=Qwen3-8B2025.12 | 9.19 | 1 | 9.19 | |
| DAPOBackbone=Qwen3-4B2025.12 | 7.39 | 1 | 7.39 | |
| DAPO-FTBackbone=Qwen3-4B2025.12 | 7.39 | 0.2 | 37.02 | |
| DAPO-FTBackbone=Qwen3-1.7B2025.12 | 5.62 | 0.2 | 28.08 | |
| DAPOBackbone=Qwen3-1.7B2025.12 | 5.46 | 1 | 5.46 | |
| GRPOBackbone=Qwen3-8B2025.12 | 4.74 | 1 | 4.74 | |
| GRPOBackbone=Qwen3-4B2025.12 | 3.37 | 1 | 3.37 | |
| INTUITORBackbone=Qwen3-8B2025.12 | 3.34 | 1 | 3.34 | |
| GRPOBackbone=Qwen3-1.7B2025.12 | 2.87 | 1 | 2.87 | |
| INTUITORBackbone=Qwen3-4B2025.12 | 2.42 | 1 | 2.42 | |
| INTUITORBackbone=Qwen3-1.7B2025.12 | 1.76 | 1 | 1.76 |