Mathematical Reasoning Evaluation on MATH
74.8UtilityPolicyAlign
Evaluation Results
| Method | Links | |
|---|---|---|
| PolicyAlignBackbone=Qwen3-4B2026.06 | 74.8 | |
| ICLBackbone=Qwen3-4B2026.06 | 74 | |
| BaseBackbone=Qwen3-4B2026.06 | 72.8 | |
| SFTBackbone=Qwen3-4B2026.06 | 70.8 |
| Method | Links | |
|---|---|---|
| PolicyAlignBackbone=Qwen3-4B2026.06 | 74.8 | |
| ICLBackbone=Qwen3-4B2026.06 | 74 | |
| BaseBackbone=Qwen3-4B2026.06 | 72.8 | |
| SFTBackbone=Qwen3-4B2026.06 | 70.8 |