List your company

Logic Puzzle on Sudoku (Success rate)

99.6Success RateActFocus
-3.98422.90849.876.692May 14, 2026

Evaluation results

MethodLinks
ActFocusAlgorithm=PPO, Model Scale=1.5B2026.0599.6
UniformAlgorithm=PPO, Model Scale=1.5B2026.0584.2
ActFocusAlgorithm=PPO, Model Scale=0.5B2026.0582
ActFocusAlgorithm=GRPO, Model Scale=1.5B2026.0580.9
UniformAlgorithm=GRPO, Model Scale=1.5B2026.0570.9
UniformAlgorithm=PPO, Model Scale=0.5B2026.0516.2
UniformAlgorithm=GRPO, Model Scale=0.5B2026.050
ActFocusAlgorithm=GRPO, Model Scale=0.5B2026.050