dataset
active
dataset:cruxeval-o-adv

cruxeval_o_adv

Code reasoning dataset with adversarially introduced errors used for reflection evaluation.

Neighborhood — ranked by edge-count