scikit-learn: Our R^2 makes odd assumptions, doesn't work with LeaveOneOut

from sklearn.datasets import make_regression
from sklearn.cross_validation import cross_val_score, LeaveOneOut
from sklearn.linear_model import Ridge

X, y, coef_ = make_regression(random_state=42, noise=1, n_samples=200, coef=True)
cross_val_score(Ridge(), X, y, cv=LeaveOneOut(len(X)))

array([ 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.])

Maybe the single sample in the LOO is not interpreted correctly? 😕