Mathematical Reasoning on AIME 25 (Pass@1, #Tokens)
43.33Pass@1 AccuracyVanilla
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| VanillaModel=R1-7B2025.12 | 43.33 | 12,139 | |
| CRESTModel=R1-7B2025.12 | 43.33 | 8,083 | |
| DynasorModel=R1-7B2025.12 | 41 | 9,360 | |
| SEALModel=R1-7B2025.12 | 36.7 | 11,692 | |
| TIPModel=R1-7B2025.12 | 33.3 | 11,225 | |
| Soft-ThinkingModel=R1-7B2025.12 | 33.3 | 11,370 | |
| CRESTModel=R1-1.5B2025.12 | 30 | 11,101 | |
| DynasorModel=R1-1.5B2025.12 | 28 | 12,412 | |
| Soft-ThinkingModel=R1-1.5B2025.12 | 23.3 | 14,843 | |
| VanillaModel=R1-1.5B2025.12 | 20 | 15,974 | |
| TIPModel=R1-1.5B2025.12 | 20 | 14,200 | |
| SEALModel=R1-1.5B2025.12 | 16.7 | 17,153 | |
| DAPO-8rollout_budget=82026.02 | 12.33 | 1,149.21 | |
| PROGRS-4rollout_budget=42026.02 | 12 | 1,065.38 | |
| PROGRS-8rollout_budget=82026.02 | 10.67 | 1,156.79 | |
| PROGRS-8 (αcoh = 0)rollout_budget=8, coherence_penalty=disabled2026.02 | 9.67 | 1,027.08 | |
| PROGRS-8 (No Centering)rollout_budget=8, centering=none2026.02 | 8.33 | 1,273.7 | |
| DAPO-16rollout_budget=162026.02 | 8 | 1,128.88 |