For a large molecular system it makes sense to converge the CDFT at a smaller basis set
first (e.g. 3-21G) and then project the MOs to the larger basis set. This is usually much
faster than applying the larger basis set right away, because the "initial guess" is way better.