Mathematical Reasoning on AMC12 (Pass@1, Pass@3)
66.27Pass@1P-ALIGN
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| P-ALIGNBackbone=Qwen3-8B2026.01 | 66.27 | 77.11 | |
| UPFT (2025)Backbone=Qwen3-8B2026.01 | 61.74 | 75.38 | |
| Zero-ShotBackbone=Qwen3-8B2026.01 | 59.84 | 71.08 | |
| SFT (Long-CoT)Backbone=Qwen3-8B2026.01 | 58.81 | 67.04 | |
| P-ALIGNBackbone=Qwen2.5-7B-Instruct2026.01 | 49.4 | 63.86 | |
| SFT (Label)Backbone=Qwen3-8B2026.01 | 48.51 | 59.04 | |
| SFT (Long-CoT)Backbone=Qwen2.5-7B-Instruct2026.01 | 48.19 | 56.63 | |
| UPFT (2025)Backbone=Qwen2.5-7B-Instruct2026.01 | 47.58 | 59.04 | |
| Zero-ShotBackbone=Qwen2.5-7B-Instruct2026.01 | 40.76 | 49.7 | |
| SFT (Label)Backbone=Qwen2.5-7B-Instruct2026.01 | 32.92 | 34.62 |