Mathematical Problem Solving on OlympiadBench (test)
42.7Pass@1 AccuracyPRM-CoT (Process-Aware)
Evaluation Results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| PRM-CoT (Process-Aware)Base Policy=Qwen2.5-Math-7B2025.12 | 42.7 | 38.42 | 52.67 | 56.67 | — | — | — | — | — | — | |
| RLVR (Ground Truth)Base Policy=Qwen2.5-Math-7B2025.12 | 40 | 36.08 | 48.67 | 54 | — | — | — | — | — | — | |
| GRPOBase Model=Qwen2.5-Math-7B, Temperature=0.6, top-p=0.95, Max tokens=4096, Scoring=exact any-boxed answer scoring2026.06 | 39.6 | 58.3 | 56.5 | 60.5 | 45.9 | 51.5 | 63.8 | 66.6 | 69.1 | 71.5 | |
| DIVER-EDModel=Qwen2.5-7B2026.06 | 39.42 | — | — | — | — | — | 71.11 | — | — | — | |
| PRM (Process-Aware)Base Policy=Qwen2.5-Math-7B2025.12 | 38.7 | 35.33 | 48.67 | 52 | — | — | — | — | — | — | |
| DIVER-TDModel=Qwen2.5-7B2026.06 | 38.36 | — | — | — | — | — | 72.19 | — | — | — | |
| REINFORCEBase Model=Qwen2.5-Math-7B, Temperature=0.6, top-p=0.95, Max tokens=4096, Scoring=exact any-boxed answer scoring2026.06 | 38.2 | 55.8 | 53.7 | 57.6 | 43.8 | 49 | 60.9 | 63.8 | 66.3 | 68.7 | |
| W-RFBase Model=Qwen2.5-Math-7B, Temperature=0.6, top-p=0.95, Max tokens=4096, Scoring=exact any-boxed answer scoring2026.06 | 38.2 | 59.4 | 57 | 61.7 | 45 | 51.3 | 65.6 | 69 | 72 | 74.6 | |
| BBGBase Model=Qwen2.5-Math-7B, Temperature=0.6, top-p=0.95, Max tokens=4096, Scoring=exact any-boxed answer scoring2026.06 | 37.1 | 61.3 | 58.8 | 64.1 | 45.1 | 52.4 | 68.3 | 72 | 75.7 | 78.5 | |
| GRPOModel=Qwen2.5-7B2026.06 | 37.08 | — | — | — | — | — | 71.7 | — | — | — | |
| SFT (Baseline)Base Policy=Qwen2.5-Math-7B2025.12 | 34.7 | 31.83 | 43.33 | 48.33 | — | — | — | — | — | — | |
| DQOModel=Qwen2.5-7B2026.06 | 31.86 | — | — | — | — | — | 66.2 | — | — | — | |
| DIVER-EDModel=Qwen2.5-3B2026.06 | 29.31 | — | — | — | — | — | 66.67 | — | — | — | |
| Base ModelBase Model=Qwen2.5-Math-7B, Temperature=0.6, top-p=0.95, Max tokens=4096, Scoring=exact any-boxed answer scoring2026.06 | 28.5 | 58 | 55.2 | 61.4 | 38.5 | 47.5 | 66.5 | 71 | 75.1 | 78.5 | |
| DIVER-TDModel=Qwen2.5-3B2026.06 | 28.22 | — | — | — | — | — | 65.93 | — | — | — | |
| GRPOModel=Qwen2.5-3B2026.06 | 27.81 | — | — | — | — | — | 64.44 | — | — | — | |
| DQOModel=Qwen2.5-3B2026.06 | 26.48 | — | — | — | — | — | 63.85 | — | — | — | |
| BaseModel=Qwen2.5-7B2026.06 | 19.1 | — | — | — | — | — | 66.52 | — | — | — | |
| BaseModel=Qwen2.5-3B2026.06 | 15.19 | — | — | — | — | — | 60.3 | — | — | — |