Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

When Emotional Stimuli meet Prompt Designing: An Auto-Prompt Graphical Paradigm

Chenggian Ma, Xiangyu Zhao, Chunhui Zhang, Yanzhao Qin, Wentao Zhang

TL;DR

The paper introduces the Auto-Prompt Graphical Paradigm (APGP), a unified approach that combines Stimulating Prompts and Framework Prompts with automated prompt filling to enhance multi-domain LLM problem solving. It defines an artifact-level workflow with immutable prompts and LLM-generated variable prompts, operationalized through $P_{desc}$, $P_{def}$, $S_1$, $S_2$, $S_3$, and $S_{Best}$, and validated via $Ans$ and $Validate$ steps. Empirical results on Ruozhiba and BBH demonstrate improved accuracy and robust reasoning across Chinese language traps, world knowledge, and logical reasoning, with ablation studies confirming the contribution of stimulating prompts. The approach paves the way for a universal, graph-based, emotion-aware prompt design paradigm that automates prompt construction while maintaining problem-solving rigor, albeit with limitations in LLM-based judgments and potential computational costs for simple tasks.

Abstract

With the development of Large Language Models (LLM), numerous prompts have been proposed, each with a rich set of features and their own merits. This paper summarizes the prompt words for large language models (LLMs), categorizing them into stimulating and framework types, and proposes an Auto-Prompt Graphical Paradigm(APGP) that combines both stimulating and framework prompts to enhance the problem-solving capabilities of LLMs across multiple domains, then exemplifies it with a framework that adheres to this paradigm. The framework involves automated prompt generation and consideration of emotion-stimulus factors, guiding LLMs in problem abstraction, diversified solutions generation, comprehensive optimization, and self-verification after providing answers, ensuring solution accuracy. Compared to traditional stimuli and framework prompts, this framework integrates the advantages of both by adopting automated approaches inspired by APE work, overcoming the limitations of manually designed prompts. Test results on the ruozhiba and BBH datasets demonstrate that this framework can effectively improve the efficiency and accuracy of LLMs in problem-solving, paving the way for new applications of LLMs.

When Emotional Stimuli meet Prompt Designing: An Auto-Prompt Graphical Paradigm

TL;DR

The paper introduces the Auto-Prompt Graphical Paradigm (APGP), a unified approach that combines Stimulating Prompts and Framework Prompts with automated prompt filling to enhance multi-domain LLM problem solving. It defines an artifact-level workflow with immutable prompts and LLM-generated variable prompts, operationalized through , , , , , and , and validated via and steps. Empirical results on Ruozhiba and BBH demonstrate improved accuracy and robust reasoning across Chinese language traps, world knowledge, and logical reasoning, with ablation studies confirming the contribution of stimulating prompts. The approach paves the way for a universal, graph-based, emotion-aware prompt design paradigm that automates prompt construction while maintaining problem-solving rigor, albeit with limitations in LLM-based judgments and potential computational costs for simple tasks.

Abstract

With the development of Large Language Models (LLM), numerous prompts have been proposed, each with a rich set of features and their own merits. This paper summarizes the prompt words for large language models (LLMs), categorizing them into stimulating and framework types, and proposes an Auto-Prompt Graphical Paradigm(APGP) that combines both stimulating and framework prompts to enhance the problem-solving capabilities of LLMs across multiple domains, then exemplifies it with a framework that adheres to this paradigm. The framework involves automated prompt generation and consideration of emotion-stimulus factors, guiding LLMs in problem abstraction, diversified solutions generation, comprehensive optimization, and self-verification after providing answers, ensuring solution accuracy. Compared to traditional stimuli and framework prompts, this framework integrates the advantages of both by adopting automated approaches inspired by APE work, overcoming the limitations of manually designed prompts. Test results on the ruozhiba and BBH datasets demonstrate that this framework can effectively improve the efficiency and accuracy of LLMs in problem-solving, paving the way for new applications of LLMs.
Paper Structure (21 sections, 4 figures, 1 table, 1 algorithm)

This paper contains 21 sections, 4 figures, 1 table, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Prompt-based LLM Reasoning
  4. Emotion-Enhanced LLM Reasoning
  5. Graph-Dependency LLM Reasoning
  6. Methodology
  7. Overview
  8. Definition
  9. Get solutions and Aggregate
  10. Get solutions and Validate
  11. Experiments
  12. Datasets
  13. Ruozhiba.
  14. BIG-Bench-Hard.
  15. Evaluation Metrics
  16. ...and 6 more sections

Figures (4)

  • Figure 2: An example for using APGP to solve a problem, which in Chinese is "How to prevent falling asleep when counting the number of sheep for the herder? ". The gradually deepening blue boxes in the picture are the answers from the AI, and the green boxes are the guiding prompt words from the auto-prompt graphical paradigm. Each pair of blue and green boxes represents an interaction with the AI, with the execution order being to execute the one on the left first and then the one on the right.
  • Figure 3: A Schematic Representation of the Stimuli Graphical Process within a Problem-Solving Framework.
  • Figure 4: Result of BIG-Bench-Hard. This result includes 23 sub-tasks in BBH, a total of 27 sub-datasets, covering multiple aspects, with the job of determining whether the output answers are correct being accomplished by LLM.
  • Figure 5: Comparison of results using Stimulating Prompts and without Stimulating Prompts on the BBH dataset.