List your company

Logical Reasoning on KK 2–3 ppl.

86.8AccuracyOracle Verifier
12.33631.6685170.332May 31, 2026

Evaluation results

MethodLinks
Oracle VerifierModel Size=12B2026.0586.8
SimpleSEModel Size=12B, Verifier threshold (τ)=0.62026.0584.8
RevisionSEModel Size=12B2026.0584.8
SimpleSEModel Size=12B, Verifier threshold (τ)=0.72026.0583
SimpleSEModel Size=12B, Verifier threshold (τ)=0.82026.0580.5
Oracle VerifierModel Size=4B2026.0578.4
SimpleSEModel Size=12B, Verifier threshold (τ)=0.52026.0578.3
gemma-3-12b-itModel Size=12B2026.0577.7
RevisionSEModel Size=4B2026.0575.8
SimpleSEModel Size=4B, Verifier threshold (τ)=0.62026.0570.9
SimpleSEModel Size=4B, Verifier threshold (τ)=0.52026.0570.8
SimpleSEModel Size=4B, Verifier threshold (τ)=0.82026.0570.4
SimpleSEModel Size=4B, Verifier threshold (τ)=0.72026.0570.1
gemma-3-4b-itModel Size=4B2026.0562
Oracle VerifierModel Size=1B2026.0532.6
SimpleSEModel Size=1B, Verifier threshold (τ)=0.82026.0523.8
RevisionSEModel Size=1B2026.0522.4
gemma-3-1b-itModel Size=1B2026.0520.9
SimpleSEModel Size=1B, Verifier threshold (τ)=0.72026.0519
SimpleSEModel Size=1B, Verifier threshold (τ)=0.62026.0517
SimpleSEModel Size=1B, Verifier threshold (τ)=0.52026.0515.2