The article discusses the significant progress in Natural Language Processing (NLP), focusing on Large Language Models (LLM) and prompt engineering techniques. The authors discuss the use of reasoning topologies like Chain-of-Thought, Tree of Thoughts, or Graph of Thoughts to guide the LLM reasoning process. They present an in-depth analysis of the prompt execution pipeline and build a taxonomy of structure-enhanced LLM reasoning schemes. The study compares existing prompting schemes, discusses design choices, performance, and cost patterns. The article also outlines theoretical underpinnings, relationships between prompting and other parts of the LLM ecosystem, and associated research challenges.
Publication date: 26 Jan 2024
Project Page: Not provided
Paper: https://arxiv.org/pdf/2401.14295
