Mathematical Problem Solving on AIME 30 samples 2025
8AccuracyHAPO
Evaluation Results
| Method | Links | |
|---|---|---|
| HAPOBackbone=LLaMA3.2-3B Instruct2025.09 | 8 | |
| HAPOBackbone=LLaMA3.1-8B-Instruct2025.09 | 1.25 | |
| Vanilla GRPOBackbone=LLaMA3.2-3B Instruct2025.09 | 0 | |
| Vanilla DAPOBackbone=LLaMA3.2-3B Instruct2025.09 | 0 | |
| DAPO w/ Forking TokensBackbone=LLaMA3.2-3B Instruct2025.09 | 0 | |
| Vanilla GRPOBackbone=LLaMA3.1-8B-Instruct2025.09 | 0 | |
| Vanilla DAPOBackbone=LLaMA3.1-8B-Instruct2025.09 | 0 | |
| DAPO w/ Forking TokensBackbone=LLaMA3.1-8B-Instruct2025.09 | 0 |