List your company
ResearchFormerly WizWand

Logical Reasoning on KK 6–8 ppl.

27.5AccuracyRevisionSE
-0.8926.47913.8521.221May 31, 2026
Updated 5mo ago

Evaluation results

MethodLinks
RevisionSEModel Size=12B2026.0527.5
Oracle VerifierModel Size=12B2026.0527
SimpleSEModel Size=12B, Verifier threshold (τ)=0.62026.0526.2
gemma-3-12b-itModel Size=12B2026.0524.4
SimpleSEModel Size=12B, Verifier threshold (τ)=0.52026.0524
SimpleSEModel Size=12B, Verifier threshold (τ)=0.72026.0524
Oracle VerifierModel Size=4B2026.0521.4
SimpleSEModel Size=12B, Verifier threshold (τ)=0.82026.0521.2
SimpleSEModel Size=4B, Verifier threshold (τ)=0.62026.0517.5
RevisionSEModel Size=4B2026.0517.1
SimpleSEModel Size=4B, Verifier threshold (τ)=0.72026.0516.4
SimpleSEModel Size=4B, Verifier threshold (τ)=0.52026.0516.3
SimpleSEModel Size=4B, Verifier threshold (τ)=0.82026.0516
gemma-3-4b-itModel Size=4B2026.0510.3
gemma-3-1b-itModel Size=1B2026.051
SimpleSEModel Size=1B, Verifier threshold (τ)=0.52026.050.8
SimpleSEModel Size=1B, Verifier threshold (τ)=0.82026.050.8
Oracle VerifierModel Size=1B2026.050.7
SimpleSEModel Size=1B, Verifier threshold (τ)=0.72026.050.4
SimpleSEModel Size=1B, Verifier threshold (τ)=0.62026.050.3
RevisionSEModel Size=1B2026.050.2