create_training_sets sets np.random.. at one point. The random seed should really be set by the main function instead.
Once we do this correctly, all existing tests should still pass if the test uses set_seed before invoking the create_training_sets function. Any other behavior is likely a bug.
create_training_sets
setsnp.random..
at one point. The random seed should really be set by themain
function instead.Once we do this correctly, all existing tests should still pass if the test uses
set_seed
before invoking thecreate_training_sets
function. Any other behavior is likely a bug.