how to get reproducible splits using LassoLarsCV in sklearn?

If I fit a LassoCV using sklearn, I can do

    model_LassoCV = LassoCV(cv=3,random_state=2018).fit(X, y)

and get reproducible splits for the cross-validation. However, if I do

    model_LassoCV = LassoLarsCV(cv=3,random_state=2018).fit(X, y)

according to the documentation, LassoLarsCV does not have a random_state. So how to get reproducible results with LassoLarsCV ?