Mathematical Reasoning on MATH500 (128-sample random subset)
71.09Top-1 AccuracyePF
Evaluation Results
| Method | Links | |
|---|---|---|
| ePFSelection=Argmax, Scoring=PRM, Base Model=Qwen2.5-7B-Instruct2025.10 | 71.09 | |
| PFSelection=Argmax, Scoring=PRM, Base Model=Qwen2.5-7B-Instruct2025.10 | 70.31 | |
| Best-of-NSelection=Argmax, Scoring=ORM, Base Model=Qwen2.5-7B-Instruct2025.10 | 67.96 | |
| ePFSelection=Argmax, Scoring=PRM, Base Model=Qwen2.5-1.5B-Instruct2025.10 | 66.42 | |
| Beam-SearchSelection=Argmax, Scoring=PRM, Base Model=Qwen2.5-7B-Instruct2025.10 | 66.4 | |
| Self-ConsistencySelection=MV, Base Model=Qwen2.5-7B-Instruct2025.10 | 65.62 | |
| Beam-SearchSelection=Argmax, Scoring=PRM, Base Model=Qwen2.5-1.5B-Instruct2025.10 | 62.5 | |
| Base SamplingBase Model=Qwen2.5-7B-Instruct2025.10 | 60.93 | |
| PFSelection=Argmax, Scoring=PRM, Base Model=Qwen2.5-1.5B-Instruct2025.10 | 60.15 | |
| Best-of-NSelection=Argmax, Scoring=ORM, Base Model=Qwen2.5-1.5B-Instruct2025.10 | 57.81 | |
| Self-ConsistencySelection=MV, Base Model=Qwen2.5-1.5B-Instruct2025.10 | 53.9 | |
| Base SamplingBase Model=Qwen2.5-1.5B-Instruct2025.10 | 45.31 |