Mathematical Reasoning on AIME 25 (Mean@16, Pass@16)
18.96Mean@16ADS
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ADSBackbone=Olmo3-7B-SFT2026.06 | 18.96 | 45.55 | |
| DOTS+RRBackbone=Olmo3-7B-SFT2026.06 | 18.47 | 44.45 | |
| ADSBackbone=Qwen3-4B-Base2026.06 | 11.53 | 37.78 | |
| GRPOBackbone=Qwen3-4B-Base2026.06 | 11.32 | 37.78 | |
| DOTS+RRBackbone=Qwen3-4B-Base2026.06 | 9.31 | 30 | |
| ADSBackbone=Qwen2.5-Math-1.5B2026.06 | 7.08 | 30 | |
| DOTS+RRBackbone=Qwen2.5-Math-1.5B2026.06 | 6.88 | 33.33 | |
| GRPOBackbone=Olmo3-7B-SFT2026.06 | 5.56 | 24.44 | |
| BaseBackbone=Olmo3-7B-SFT2026.06 | 5.42 | 27.78 | |
| BaseBackbone=Qwen3-4B-Base2026.06 | 4.23 | 27.78 | |
| GRPOBackbone=Qwen2.5-Math-1.5B2026.06 | 3.47 | 20 | |
| BaseBackbone=Qwen2.5-Math-1.5B2026.06 | 1.74 | 13.33 |