List your company

Rubric Discovery on Arena-Expert-5K, HelpSteer3, HH-RLHF, UltraFeedback cross-source mean

4.45St. ScorePREMISE (+ VFR unconstrained)
2.02682.65593.2853.9141May 29, 2026

Evaluation results

MethodLinks
PREMISE (+ VFR unconstrained)K=18.02026.054.4548.831.358.529.955.20.091464.866.489.9
CRITIQK=5.82026.054.435.834.659.243.975.30.01768.169.587.8
PREMISE (+ VFR constrained)K=18.02026.054.3951.927.460.93657.70.091464.96790.9
PREMISE (discovered)K=20.02026.054.2153.126.359.746.457.70.08156566.591.9
AUTO-RUBRICK=20.02026.053.9431.913.936.529.730.021068.763.790.8
PREMISE (+ pref repair)K=20.02026.053.8349.924.154.843.2340.081168.668.591.6
INVERSE CAIK=2.82026.053.4455.837.854.451.7630.05467.561.493.2
AGENTEVALK=4.02026.053.3858.228.170.23382.40.06368.266.587.3
AUTORULEK=26.22026.052.1251.325.958.550.240.30.041868.265.788.6