TL;DRAMPLIFY 框架使用后续解释的方法,自动生成自然语言解释以提供纠正信号,从而提高 Large Language Models 的预测准确率。
Abstract
large language models (LLMs) have demonstrated remarkable capabilities in
performing complex tasks. Moreover, recent research has shown that
incorporating human-annotated rationales (e.g., Chain-of- Thought prompting)
during in-context learning can significantly enhance the performance