List your company

Math Reasoning on AIME 24 (Pass@1, #Tokens)

73.3Pass@1 ScoreDynasor-CoT
4.03622.0184057.982Mar 12, 2026

Evaluation results

MethodLinks
Dynasor-CoTBackbone=Qwen3-14B2026.0373.310,369
ReBalanceBackbone=Qwen3-14B2026.0373.39,464
NoThinkingBackbone=Qwen3-14B2026.03708,898
DEERBackbone=QwQ-32B2026.03708,885
ReBalanceBackbone=QwQ-32B2026.037010,350
BaselineBackbone=Qwen3-14B2026.0366.710,888
CoDBackbone=Qwen3-14B2026.0366.710,212
DEERBackbone=Qwen3-14B2026.0366.79,973
BaselineBackbone=QwQ-32B2026.0366.714,342
NoThinkingBackbone=QwQ-32B2026.0366.710,507
NoWaitBackbone=QwQ-32B2026.0366.78,190
SEALBackbone=Qwen3-14B2026.0363.310,322
CoDBackbone=QwQ-32B2026.0363.311,438
Dynasor-CoTBackbone=QwQ-32B2026.0363.311,156
SEALBackbone=QwQ-32B2026.0363.310,344
NoWaitBackbone=Qwen3-14B2026.036010,507
FlashThinkBackbone=QwQ-32B2026.036010,034
TrimRBackbone=QwQ-32B2026.0356.78,345
Manifold SteeringBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0353.38,457
ReBalanceBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0353.39,948
DEERBackbone=DeepSeek-R1-Distill-Qwen-7B2026.03508,924
NoWaitBackbone=DeepSeek-R1-Distill-Qwen-7B2026.03506,844
BaselineBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0346.711,923
CoDBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0346.711,663
Dynasor-CoTBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0346.79,864
SEALBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0343.310,112
CoDBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.0333.311,894
ReBalanceBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.0333.310,668
NoWaitBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.03308,225
Manifold SteeringBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.033010,134
Dynasor-CoTBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.0326.710,564
NoThinkingBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0326.74,427
BaselineBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.0323.312,451
SEALBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.0323.310,785
DEERBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.03208,135
NoThinkingBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.036.76,998