Unless I misread your question, your problem is identical to maximum cut when $G'=G$, implying your problem is NP-hard.
(Actually, as David Benson-Putnins points out, I did misread the question. My answer requires the edge weights to be not necessarily positive.)