ResearchBenchmarksLogical Reasoning on FOLIO n=203 (held-out)Follow0.7195Per-Class LossΠ10.7180960.7275730.737050.746527May 7, 2026Evaluation ResultsMethodMethodLinksPer-Class LossΠ1Controller Class=Π1 (K...Controller Class=Π1 (KMeans-K=6), CV protocol=Strict nested 5-fold-by-5-seed2026.050.7195Π0Controller Class=Π0 (d...Controller Class=Π0 (direct policy), CV protocol=Strict nested 5-fold-by-5-seed2026.050.752Π2Controller Class=Π2, C...Controller Class=Π2, CV protocol=Strict nested 5-fold-by-5-seed2026.050.7546