Mathematical Reasoning on AMC23 (Avg@16, Avg)
90.62Avg@16OISD
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| OISDBackbone=Qwen3-8B2026.05 | 90.62 | 66.72 | |
| BuPOBackbone=Qwen3-8B2026.05 | 89.22 | 66.36 | |
| GRPOBackbone=Qwen3-8B2026.05 | 85.94 | 64.23 | |
| OISDBackbone=Qwen3-4B2026.05 | 84.38 | 64.75 | |
| BuPOBackbone=Qwen3-4B2026.05 | 81.09 | 58.51 | |
| RLOOBackbone=Qwen3-4B2026.05 | 77.66 | 54 | |
| PPOBackbone=Qwen3-4B2026.05 | 77.03 | 55.22 | |
| GRPOBackbone=Qwen3-4B2026.05 | 76.88 | 55.08 | |
| VanillaBackbone=Qwen3-4B2026.05 | 67.66 | 47.44 | |
| VanillaBackbone=Qwen3-8B2026.05 | 67.34 | 48.49 | |
| Reinforce++Backbone=Qwen3-4B2026.05 | 63.44 | 45.03 | |
| OISDBackbone=OctoThinker-8B-Long-Base2026.05 | 39.38 | 30.09 | |
| BuPOBackbone=OctoThinker-8B-Long-Base2026.05 | 37.66 | 27.79 | |
| GRPOBackbone=OctoThinker-8B-Long-Base2026.05 | 34.84 | 24.11 | |
| OISDBackbone=OctoThinker-3B-Long-Base2026.05 | 27.81 | 19.91 | |
| GRPOBackbone=OctoThinker-3B-Long-Base2026.05 | 27.5 | 18.58 | |
| BuPOBackbone=OctoThinker-3B-Long-Base2026.05 | 27.5 | 19.59 | |
| VanillaBackbone=OctoThinker-8B-Long-Base2026.05 | 4.53 | 3.75 | |
| VanillaBackbone=OctoThinker-3B-Long-Base2026.05 | 1.24 | 1.68 |