NeuRes: Learning Proofs of Propositional Satisfiability