Mathematical Reasoning on AIME 25 (pass@1 mean/std)
48.3Pass@1 Mean ScoreGRPO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GRPOSampling strategy=avg@4, Number of repeats=42026.01 | 48.3 | 9.1 | |
| ReNCESampling strategy=avg@4, Number of repeats=42026.01 | 43.3 | 9.1 | |
| DAPOSampling strategy=avg@4, Number of repeats=42026.01 | 35.8 | 7.7 | |
| Semi-online DPOSampling strategy=avg@4, Number of repeats=42026.01 | 31.7 | 9.1 | |
| No trainingSampling strategy=avg@4, Number of repeats=42026.01 | 19.2 | 6 |