Math Reasoning on Olympiad (Pass@1, Tokens)
68.6Pass@1ReBalance
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ReBalanceBackbone=QwQ-32B2026.03 | 68.6 | 7,422 | |
| CoDBackbone=QwQ-32B2026.03 | 67.7 | 7,028 | |
| SEALBackbone=QwQ-32B2026.03 | 67.5 | 7,371 | |
| BaselineBackbone=QwQ-32B2026.03 | 66.7 | 8,219 | |
| ReBalanceBackbone=Qwen3-14B2026.03 | 66.3 | 7,257 | |
| DEERBackbone=Qwen3-14B2026.03 | 66.1 | 6,849 | |
| NoThinkingBackbone=QwQ-32B2026.03 | 66.1 | 8,132 | |
| DEERBackbone=QwQ-32B2026.03 | 64.3 | 7,085 | |
| NoThinkingBackbone=Qwen3-14B2026.03 | 64 | 5,880 | |
| NoWaitBackbone=QwQ-32B2026.03 | 62.6 | 8,223 | |
| SEALBackbone=Qwen3-14B2026.03 | 62.3 | 7,131 | |
| CoDBackbone=Qwen3-14B2026.03 | 62.2 | 6,554 | |
| BaselineBackbone=Qwen3-14B2026.03 | 60.6 | 7,450 | |
| NoWaitBackbone=Qwen3-14B2026.03 | 59.2 | 7,332 | |
| ReBalanceBackbone=DeepSeek-R1-Distill-Qwen-7B2026.03 | 57 | 6,321 | |
| BaselineBackbone=DeepSeek-R1-Distill-Qwen-7B2026.03 | 56.1 | 7,590 | |
| Dynasor-CoTBackbone=DeepSeek-R1-Distill-Qwen-7B2026.03 | 55.4 | 7,427 | |
| DEERBackbone=DeepSeek-R1-Distill-Qwen-7B2026.03 | 53.9 | 5,804 | |
| SEALBackbone=DeepSeek-R1-Distill-Qwen-7B2026.03 | 53.9 | 6,261 | |
| NoWaitBackbone=DeepSeek-R1-Distill-Qwen-7B2026.03 | 52.1 | 4,760 | |
| CoDBackbone=DeepSeek-R1-Distill-Qwen-7B2026.03 | 47.5 | 5,688 | |
| NoThinkingBackbone=DeepSeek-R1-Distill-Qwen-7B2026.03 | 45.3 | 3,331 | |
| ReBalanceBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.03 | 43.9 | 7,235 | |
| Dynasor-CoTBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.03 | 42.6 | 8,859 | |
| BaselineBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.03 | 41.2 | 8,785 | |
| NoWaitBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.03 | 40.6 | 4,795 | |
| NoThinkingBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.03 | 37.3 | 3,485 | |
| DEERBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.03 | 35.4 | 5,982 | |
| CoDBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.03 | 35.2 | 7,160 | |
| SEALBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.03 | 32.7 | 7,117 |