Refining Sentence Embedding Model through Ranking Sentences Generation with Large Language Models