Adversarial attacks and defenses for large language models (LLMs): methods, frameworks & challenges