Large language models (LLMs) demonstrated impressive few-shot learning capabilities, rapidly adapting to latest tasks with only a handful of examples.Nonetheless, despite their advances, LLMs still face limitations in complex reasoning involving chaotic contexts overloaded...