Mathematical Reasoning on AMC 23 (Pass@16, Mean@16)
55.57Mean@16ADS
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ADSBackbone=Olmo3-7B-SFT2026.06 | 55.57 | 95 | |
| ADSBackbone=Qwen3-4B-Base2026.06 | 54.89 | 89.17 | |
| GRPOBackbone=Qwen3-4B-Base2026.06 | 53.65 | 89.17 | |
| DOTS+RRBackbone=Qwen3-4B-Base2026.06 | 52.91 | 89.17 | |
| DOTS+RRBackbone=Olmo3-7B-SFT2026.06 | 49.22 | 85.83 | |
| ADSBackbone=Qwen2.5-Math-1.5B2026.06 | 45.89 | 87.5 | |
| DOTS+RRBackbone=Qwen2.5-Math-1.5B2026.06 | 44.89 | 83.33 | |
| GRPOBackbone=Olmo3-7B-SFT2026.06 | 40.47 | 84.17 | |
| GRPOBackbone=Qwen2.5-Math-1.5B2026.06 | 35.68 | 79.17 | |
| BaseBackbone=Olmo3-7B-SFT2026.06 | 34.9 | 80.83 | |
| BaseBackbone=Qwen3-4B-Base2026.06 | 23.7 | 78.33 | |
| BaseBackbone=Qwen2.5-Math-1.5B2026.06 | 22.45 | 80.83 |