ResearchBenchmarksSeizure burden minimization on Intrinsic (held-out)Follow0.4985Worst RewardBaseline0.3280440.3722970.416550.460803Jun 3, 2026Evaluation ResultsMethodMethodLinksWorst RewardMean RewardWorst-Case GainBaselineCondition=Baseline, n=8Condition=Baseline, n=82026.060.49850.4602—OptimizedCondition=Optimized, n=8Condition=Optimized, n=82026.060.33460.284832.9