Mathematical Reasoning on AMC23 (Avg@16 and Pass@16)
53.75Average Score @16PriFT-mass
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| PriFT-massBackbone=Qwen2.5-Math (7B)2026.06 | 53.75 | 90 | |
| PriFT-probBackbone=Qwen2.5-Math (7B)2026.06 | 49.84 | 90 | |
| PriFT-massBackbone=Qwen3-Base (8B)2026.06 | 47.03 | 85 | |
| IDFTBackbone=Qwen2.5-Math (7B)2026.06 | 45.31 | 77.5 | |
| PriFT-probBackbone=Qwen3-Base (8B)2026.06 | 44.06 | 82.5 | |
| DFTBackbone=Qwen3-Base (8B)2026.06 | 43.59 | 77.5 | |
| IDFTBackbone=Qwen3-Base (8B)2026.06 | 43 | 67.5 | |
| ASFTBackbone=Qwen2.5-Math (7B)2026.06 | 42.5 | 77.5 | |
| TALRBackbone=Qwen2.5-Math (7B)2026.06 | 42.03 | 80 | |
| DFTBackbone=Qwen2.5-Math (7B)2026.06 | 40.94 | 72.5 | |
| TALRBackbone=Qwen3-Base (8B)2026.06 | 40.9 | 82.5 | |
| ASFTBackbone=Qwen3-Base (8B)2026.06 | 39.06 | 85 | |
| OriginalBackbone=Qwen3-Base (8B)2026.06 | 35.47 | 85 | |
| SFTBackbone=Qwen2.5-Math (7B)2026.06 | 25 | 70 | |
| OriginalBackbone=Qwen2.5-Math (7B)2026.06 | 24.84 | 82.5 | |
| EAFTBackbone=Qwen3-Base (8B)2026.06 | 24.2 | 70 | |
| SFTBackbone=Qwen3-Base (8B)2026.06 | 22.81 | 62.5 | |
| EAFTBackbone=Qwen2.5-Math (7B)2026.06 | 22.19 | 75 |