Mathematical Reasoning on DEEPMATH 128 samples
35.93Top-1 AccuracyePF
Evaluation Results
| Method | Links | |
|---|---|---|
| ePFSelection=Argmax, Scoring=PRM, Base Model=Qwen2.5-7B-Instruct2025.10 | 35.93 | |
| PFSelection=Argmax, Scoring=PRM, Base Model=Qwen2.5-7B-Instruct2025.10 | 34.37 | |
| Best-of-NSelection=Argmax, Scoring=ORM, Base Model=Qwen2.5-7B-Instruct2025.10 | 32.03 | |
| Beam-SearchSelection=Argmax, Scoring=PRM, Base Model=Qwen2.5-7B-Instruct2025.10 | 32.03 | |
| Self-ConsistencySelection=MV, Base Model=Qwen2.5-7B-Instruct2025.10 | 30.46 | |
| ePFSelection=Argmax, Scoring=PRM, Base Model=Qwen2.5-1.5B-Instruct2025.10 | 25 | |
| Base SamplingBase Model=Qwen2.5-7B-Instruct2025.10 | 23.43 | |
| PFSelection=Argmax, Scoring=PRM, Base Model=Qwen2.5-1.5B-Instruct2025.10 | 22.65 | |
| Beam-SearchSelection=Argmax, Scoring=PRM, Base Model=Qwen2.5-1.5B-Instruct2025.10 | 21.09 | |
| Best-of-NSelection=Argmax, Scoring=ORM, Base Model=Qwen2.5-1.5B-Instruct2025.10 | 20.31 | |
| Self-ConsistencySelection=MV, Base Model=Qwen2.5-1.5B-Instruct2025.10 | 13.28 | |
| Base SamplingBase Model=Qwen2.5-1.5B-Instruct2025.10 | 10.15 |