Did arm succeed on EVERY trial for metric — τ-bench pass^k = 1? The reliability predicate over an eval report (vs. improvement, which reads the mean gap). Reads report.arms[arm].stats[metric].passK.
arm
metric
improvement
report.arms[arm].stats[metric].passK
Did
armsucceed on EVERY trial formetric— τ-bench pass^k = 1? The reliability predicate over an eval report (vs.improvement, which reads the mean gap). Readsreport.arms[arm].stats[metric].passK.