Mathematical Reasoning on MATH 500 (Pass@16, Mean@16)
67.56Mean@16ADS
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ADSBackbone=Olmo3-7B-SFT2026.06 | 67.56 | 88.2 | |
| ADSBackbone=Qwen3-4B-Base2026.06 | 67.13 | 84.73 | |
| DOTS+RRBackbone=Olmo3-7B-SFT2026.06 | 64.23 | 86.27 | |
| DOTS+RRBackbone=Qwen3-4B-Base2026.06 | 63.85 | 84 | |
| ADSBackbone=Qwen2.5-Math-1.5B2026.06 | 63.47 | 84.8 | |
| DOTS+RRBackbone=Qwen2.5-Math-1.5B2026.06 | 61.68 | 83.33 | |
| GRPOBackbone=Qwen3-4B-Base2026.06 | 61.49 | 84.73 | |
| GRPOBackbone=Olmo3-7B-SFT2026.06 | 59.35 | 84.27 | |
| GRPOBackbone=Qwen2.5-Math-1.5B2026.06 | 55.04 | 82.87 | |
| BaseBackbone=Olmo3-7B-SFT2026.06 | 54.71 | 82.67 | |
| BaseBackbone=Qwen2.5-Math-1.5B2026.06 | 38.46 | 80.6 | |
| BaseBackbone=Qwen3-4B-Base2026.06 | 32.15 | 79.87 |