large language models have manifested remarkable capabilities by leveraging
chain-of-thought (CoT) reasoning techniques to solve intricate questions
through step-by-step reasoning chains. Despite its success, the efficacy of
such reasoning is inherently contingent upon the quality of C