Are gradient information useful for pruning of LLMs? - View it on GitHub