Mathematical Problem Solving on AMC 83 samples
59.32Accuracy (AMC 83)HAPO
Evaluation Results
| Method | Links | |
|---|---|---|
| HAPOBackbone=LLaMA3.1-8B-Instruct2025.09 | 59.32 | |
| HAPOBackbone=LLaMA3.2-3B Instruct2025.09 | 58.74 | |
| DAPO w/ Forking TokensBackbone=LLaMA3.1-8B-Instruct2025.09 | 58.46 | |
| Vanilla DAPOBackbone=LLaMA3.1-8B-Instruct2025.09 | 57.11 | |
| DAPO w/ Forking TokensBackbone=LLaMA3.2-3B Instruct2025.09 | 55.96 | |
| Vanilla GRPOBackbone=LLaMA3.1-8B-Instruct2025.09 | 54.16 | |
| Vanilla DAPOBackbone=LLaMA3.2-3B Instruct2025.09 | 52.11 | |
| Vanilla GRPOBackbone=LLaMA3.2-3B Instruct2025.09 | 49.28 |