Mathematical Reasoning on AIME 25 (Best AIME25)
58.3Best AIME25 ScoreGuided OPSD
Evaluation Results
| Method | Links | |
|---|---|---|
| Guided OPSDPersona source=system2026.05 | 58.3 | |
| EDGE-OPDAblation=w/o KL, Persona source=system2026.05 | 57.5 | |
| RLSD-no-verifierGuidance=guided2026.05 | 56.9 | |
| EDGE-OPDPersona source=user2026.05 | 55.6 | |
| RLSD-no-verifierGuidance=unguided2026.05 | 55 | |
| EDGE-OPDAblation=w/o KL, Persona source=user2026.05 | 55 | |
| Guided OPSDPersona source=user2026.05 | 54.4 | |
| EDGE-OPDPersona source=system2026.05 | 53.3 | |
| Nemotron-3-Nano-4BDescription=Base model2026.05 | 53.1 | |
| OPSDGuidance=none2026.05 | 51.7 |