Mathematical Reasoning on GSM8K (Composite Score)
88.16Composite ScoreSFTKey-Tag
Evaluation Results
| Method | Links | |
|---|---|---|
| SFTKey-TagModel=Qwen3-8B-Base2025.12 | 88.16 | |
| SFT-TagModel=Qwen3-8B-Base2025.12 | 88.05 | |
| SFTKey-TagModel=Qwen2.5-7B2025.12 | 84.2 | |
| SFT(baseline)Model=Qwen2.5-7B2025.12 | 83.05 | |
| SFT-TagModel=Qwen2.5-7B2025.12 | 83.02 | |
| SFT(baseline)Model=Qwen3-8B-Base2025.12 | 82.18 | |
| SFTKey-TagModel=SmolLM3-3B-Base2025.12 | 77.98 | |
| SFT-TagModel=SmolLM3-3B-Base2025.12 | 77.75 | |
| SFT(baseline)Model=SmolLM3-3B-Base2025.12 | 76.85 | |
| Key-TagModel=Qwen3-8B-Base2025.12 | 73.6 | |
| Key-TagModel=SmolLM3-3B-Base2025.12 | 58.5 | |
| Key-TagModel=Qwen2.5-7B2025.12 | 55.19 | |
| Key-TagModel=Qwen2.5-3B2025.12 | 50.3 | |
| SFT(baseline)Model=Qwen2.5-3B2025.12 | 47.49 | |
| SFTKey-TagModel=Qwen2.5-3B2025.12 | 47.39 | |
| SFT-TagModel=Qwen2.5-3B2025.12 | 38.66 | |
| SFTKey-TagModel=Qwen2.5-1.5B2025.12 | 36.88 | |
| Key-TagModel=Qwen2.5-1.5B2025.12 | 34.3 | |
| SFT-TagModel=Qwen2.5-1.5B2025.12 | 33.32 | |
| SFT(baseline)Model=Qwen2.5-1.5B2025.12 | 32.48 |