Unlocking The Power Of Self-rewarding Language Models!