Logical Reasoning on Logic-RL
92.69Average ScoreHAPO
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| HAPOModel=Qwen2.5-7B-Instruct-1M2025.09 | 92.69 | 95.41 | 93.34 | 93.26 | 88.74 | |
| DAPO w/ Forking TokensModel=Qwen2.5-7B-Instruct-1M2025.09 | 90.15 | 93.44 | 92.13 | 90.26 | 84.75 | |
| EDGE-GRPOModel=Qwen2.5-7B-Instruct-1M2025.09 | 88.92 | 92.35 | 91.13 | 89.58 | 82.63 | |
| ArcherModel=Qwen2.5-7B-Instruct-1M2025.09 | 87.68 | 94.12 | 89.75 | 85.28 | 81.55 | |
| Vanilla DAPOModel=Qwen2.5-7B-Instruct-1M2025.09 | 87.13 | 92.23 | 90.16 | 85.72 | 80.42 |