List your company

Multi-domain reasoning on MMLU-Pro

88.7Math AccuracySCVC
65.19671.29877.483.502Mar 19, 2026

Evaluation results

MethodLinks
SCVCSampling budget (K)=82026.0388.7—77.981.786.278.783.878.787.378.668.186.171.678.279.7—
SCVCSampling budget (K)=52026.0387.9—77.48185.677.983.378.18777.767.485.670.977.879.1—
SCVCSampling budget (K)=22026.0385.5—76.279.483.775.982.377.486.575.565.484.369.376.977.8—
VCSampling budget (K)=82026.0384.9—74.377.18374.178.6768373.759.483.267.575.975.7—
VCSampling budget (K)=52026.0384.5—73.876.682.673.778.375.782.572.959.68367.375.575.3—
VCSampling budget (K)=22026.0381.1—71.974.580.271.876.174.280.970.458.581.265.473.773.2—
VCSampling budget (K)=12026.0378.7—69.972.47869.774.67279.268.157.479.263.671.771.3—
SCSampling budget (K)=82026.0377.6—71.874.578.172.679.468.883.572.566.876.366.671.773.9—
SCSampling budget (K)=52026.0374.1—69.571.775.3707766.781.97065.573.764.869.971.7—
SCSampling budget (K)=22026.0366.1—63.464.867.463.270.662.374.663.361.466.36063.565.2—