List your company

Combinatorial Reasoning on 3-SAT

90.9AccuracyPolicy
75.379.3583.487.45May 29, 2026

Evaluation results

MethodLinks
PolicyTraining protocol=Joint training2026.0590.9
High conf.Training protocol=Joint training2026.0589.8
BaselineTraining protocol=Joint training2026.0588.8
MarginTraining protocol=Joint training2026.0585.2
Oracle policyTraining protocol=Policy-only training2026.0582.3
PolicyTraining protocol=Policy-only training2026.0576.1
MarginTraining protocol=Policy-only training2026.0576
High conf.Training protocol=Policy-only training2026.0575.9