Mathematical Problem Solving on MATH 500 (Accuracy, Refinement Gap)
93.5AccuracyBoN@4
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| BoN@4backbone=GSR-7B, Aggregation Model=RM-R1-7B2025.08 | 93.5 | — | |
| selfRef@4backbone=GSR-7B2025.08 | 93.4 | 0.6 | |
| maj@5backbone=GSR-7B2025.08 | 93.1 | — | |
| maj@4backbone=GSR-7B2025.08 | 92.8 | — | |
| BoN@5backbone=GSR-7B, Aggregation Model=SkyworkRM-27B2025.08 | 92.7 | — | |
| BoN@4backbone=GSR-7B, Aggregation Model=SkyworkRM-27B2025.08 | 92.5 | — | |
| Ref@4backbone=GSR-7B, Aggregation Model=Synthesizer-8B2025.08 | 92 | — | |
| QwQ-Previewmode=pass@12025.08 | 90.6 | — | |
| pass@1backbone=GSR-7B2025.08 | 90.6 | — | |
| BoN@4backbone=GSR-7B, Aggregation Model=RRM-7B2025.08 | 90.6 | — | |
| o1-minimode=pass@12025.08 | 90 | — | |
| maj@4backbone=Qwen2.5-7B-Instruct (Base)2025.08 | 79.4 | — | |
| selfRef@4backbone=Qwen2.5-7B-Instruct (Base)2025.08 | 78.7 | -0.7 | |
| BoN@4backbone=Qwen2.5-7B-Instruct (Base), Aggregation Model=SkyworkRM-27B2025.08 | 78.6 | — | |
| pass@1backbone=Qwen2.5-7B-Instruct (Base)2025.08 | 75.9 | — |