Maj@k: strict-majority correctness over k samples — probability that a uniformly sampled subset of k observed traces contains strictly more than half correct solutions. Equivalent to G-Pass@k_τ with τ = (⌊k/2⌋+1)/k.
k
Maj@k: strict-majority correctness over
ksamples — probability that a uniformly sampled subset ofkobserved traces contains strictly more than half correct solutions. Equivalent to G-Pass@k_τ with τ = (⌊k/2⌋+1)/k.