List your company

Mathematical Reasoning on MATH (Accuracy, Length)

93.67AccuracyAutoTHINK
68.7175.1981.6788.15Feb 2, 2026

Evaluation results

MethodLinks
AutoTHINKBackbone=DeepSeek-R1-7B2026.0293.672,449.16
StandardpBackbone=DeepSeek-R1-7B2026.0292.334,261.18
C3oTBackbone=DeepSeek-R1-7B2026.02923,805.73
S3-CoTBackbone=DeepSeek-R1-7B2026.02922,833.27
EffReaBackbone=DeepSeek-R1-7B2026.0291.332,700.98
LASERDEBackbone=DeepSeek-R1-7B2026.0291.331,795.42
S3-CoTscBackbone=DeepSeek-R1-7B2026.0290.673,016.64
EfficientpBackbone=DeepSeek-R1-7B2026.0290.333,206.63
LC-R1Backbone=DeepSeek-R1-7B2026.0288.671,374.06
CoT-ValveBackbone=DeepSeek-R1-7B2026.0280.331,499.4
TokenSkipBackbone=Qwen2.5-7B2026.0274.67512.12
StandardpBackbone=Qwen2.5-7B2026.0272.67559.49
ShorterBetterBackbone=DeepSeek-R1-7B2026.0272.33585.57
C3oTBackbone=Qwen2.5-7B2026.0271.33536.77
CoT-ValveBackbone=Qwen2.5-7B2026.0271619.19
S3-CoTBackbone=Qwen2.5-7B2026.0270.5426.91
EfficientpBackbone=Qwen2.5-7B2026.0270300.83
S3-CoTscBackbone=Qwen2.5-7B2026.0269.67433.43